Citation Summary
| Citing sentences |
|---|
| N07-3002 1 10:74 Subsequent research began to focus more on conditional models of parse structure given the input sentence, which allowed discriminative training techniques such as maximum conditional likelihood (i.e. maximum entropy ) to be applied (Ratnaparkhi, 1999; Charniak, 2000). |
| N07-3002 2 7:74 1 Introduction Over the past decade, there has been tremendous progress on learning parsing models from treebank data (Magerman, 1995; Collins, 1999; Charniak, 1997; Ratnaparkhi, 1999; Charniak, 2000; Wang et al. , 2005; McDonald et al. , 2005). |
| W06-2303 3 140:147 These systems use (Charniak, 2000)s parse trees both for training and testing as well as various other information sources including sets of n-best parse trees (Punyakanok et al. , 2005; Haghighi et al. , 2005) or chunks (Marquez et al. , 2005; Pradhan et al. , 2005) and named entities (Surdeanu and Turmo, 2005). |
| W06-2303 4 3:147 1 Introduction Recent successes in statistical syntactic parsing based on supervised learning techniques trained on a large corpus of syntactic trees (Collins, 1999; Charniak, 2000; Henderson, 2003) have brought forth the hope that the same approaches could be applied to the more ambitious goal of recovering the propositional content and the frame semantics of a sentence. |
| W06-2303 5 139:147 The partial trees output by these systems were merged with the parse trees returned by (Charniak, 2000)s parser. |
| W06-2303 6 39:147 As with many other statistical parsers (Collins, 1999; Charniak, 2000), SSN parsers use a history-based model of parsing. |
| W06-2303 7 135:147 However, state-ofthe-art semantic role labelling systems (CoNLL, 2005) use parse trees output by state-of-the-art parsers (Collins, 1999; Charniak, 2000), both for training and testing, and return partial trees annotated with semantic role labels. |
| P06-1006 8 185:193 We can see that Charniak (2000)s parser leads to higher success rates for NPaper and BNews, while Collins (1999)s achieves better results for NWire. |
| P06-1006 9 121:193 The texts were parsed using the maximum-entropybased Charniak parser (Charniak, 2000), based on which the structured features were computed automatically. |
| P06-1006 10 182:193 5.6 Comparison with Different Parsers As mentioned, the above reported results were based on Charniak (2000)s parser. |
| P06-2067 11 180:193 All these results seem to suggest that adding punctuation in speech transcription is of little help to statistical parsers including at least three state-of-the-art statistical parsers (Collins, 1999; Charniak, 2000; Bikel, 2004). |
| P06-2067 12 8:193 1 Introduction Robust statistical syntactic parsers, made possible by new statistical techniques (Collins, 1999; Charniak, 2000; Bikel, 2004) and by the availability of large, hand-annotated training corpora such as WSJ (Marcus et al. , 1993) and Switchboard (Godefrey et al. , 1992), have had a major impact on the field of natural language processing. |
| W04-2506 13 39:256 To determine which word indicates the semantic class of the expected answer, the syntactic dependencies1 between the question words may be employed (Harabagiu 1Syntactic parsers publicly available, e.g., (Charniak, 2000; et al. , 2000; Pasca and Harabagiu, 2001; Harabagiu et al. , 2001). |
| P06-1051 14 160:203 We also used the following resources: The Charniak parser (Charniak, 2000) and the morphalemmatiser (Minnen et al. , 2001) to carry out the syntactic and morphological analysis. |
| W03-1210 15 88:174 Many researchers ((Blaheta and Charniak 2000), (Gildea and Jurafsky 2000), showed that lexical and syntactic information is very useful for predicateargument recognition tasks, such as semantic roles. |
| N06-1039 16 129:236 3.2 Parsing and GLARFing After getting a set of basic clusters, we pass them to an existing statistical parser (Charniak, 2000) and rule-based tree normalizer to obtain a GLARF structure for each sentence in every article. |
| W04-2003 17 45:197 Statistical disambiguation such as (Collins and Brooks, 1995) for PP-attachment or (Collins, 1997; Charniak, 2000) for generative parsing greatly improve disambiguation, but as they model by imitation instead of by understanding, complete soundness has to remain elusive. |
| W04-2003 18 106:197 Much of the interesting work is determining what goes into [the history] H(c)(Charniak, 2000). |
| W04-2003 19 28:197 A number of robust statistical parsers that oer solutions to these problems have now become available (Charniak, 2000; Collins, 1999; Henderson, 2003), but they typically produce CFG constituency data as output, trees that do not express long-distance dependencies. |
| W06-1601 20 132:193 5 Datasets and Evaluation We train our models with verb instances extracted from three parsed corpora: (1) the Wall Street Journal section of the Penn Treebank (PTB), which was parsed by human annotators (Marcus et al. , 1993), (2) the Brown Laboratory for Linguistic Information Processing corpus of Wall Street Journal text (BLLIP), which was parsed automatically by the Charniak parser (Charniak, 2000), and (3) the Gigaword corpus of raw newswire text (GW), which we parsed ourselves with the Stanford parser. |
| P03-1013 21 177:266 Table 4 shows the linguistic features of the resulting model compared to the models of Carroll and Rooth (1998), Collins (1997), and Charniak (2000). |
| P03-1013 22 222:266 TnT tagging Perfect tagging LR LP LR LP PP 3.45 1.60 4.21 3.35 S 1.28 0.11 2.23 1.22 Coord 1.87 0.39 1.54 0.80 VP 0.72 0.18 0.58 0.30 AP 0.57 0.10 0.08 0.07 AV P 0.32 0.44 0.10 0.11 NP 0.06 0.78 0.15 0.02 Table 6: Change in performance when reverting to head-head statistics for individual categories ter information (Charniak, 2000), as illustrated in Table 4. |
| P03-1013 23 171:266 sister head tag X Table 4: Linguistic features in the current model compared to the models of Carroll and Rooth (1998), Collins (1997), and Charniak (2000) Negra, based on Collinss (1997) model for nonrecursive NPs in the Penn Treebank (which are also flat). |
| P03-1013 24 12:266 Lexicalization can increase parsing performance dramatically for English (Carroll and Rooth, 1998; Charniak, 1997, 2000; Collins, 1997), and the lexicalized model proposed by Collins (1997) has been successfully applied to Czech (Collins et al. , 1999) and Chinese (Bikel and Chiang, 2000). |
| P03-1013 25 55:266 3 Probabilistic Parsing Models 3.1 Probabilistic Context-Free Grammars Lexicalization has been shown to improve parsing performance for the Penn Treebank (e.g. , Carroll and Rooth 1998; Charniak 1997, 2000; Collins 1997). |
| P03-1013 26 7:266 1 Introduction Treebank-based probabilistic parsing has been the subject of intensive research over the past few years, resulting in parsing models that achieve both broad coverage and high parsing accuracy (e.g. , Collins 1997; Charniak 2000). |
| P03-1013 27 180:266 Charniaks (2000) model extends this to higher order Markov chains (first to third order), and therefore includes category information about previous sisters.The current model differs from all these proposals: it does not use any information about the head sister, but instead includes the category, head word, and head tag of the previous sister, effectively treating it as the head. |
| P07-1104 28 131:188 The results are presented in Tables 12 and 2, where Baseline results were obtained using the parser by Charniak (2000). |
| P07-1104 29 124:188 The linear model combines the log probability calculated by the Charniak (2000) parser as a feature with 1,219,272 additional features. |
| P07-1104 30 24:188 For example, in the parse re-ranking task, one cannot tell whether the L2regularized ME approach used by Charniak and Johnson (2005) significantly outperforms the Boosting method by Collins (2000) because different feature sets and n-best parses were used in the evaluations of these methods. |
| P07-1104 31 15:188 ME estimators with L2 regularization, which have been widely used in NLP tasks (e.g. , Chen and Rosenfeld 2000; Charniak and Johnson 2005; Johnson et al. 1999), tend to produce models that have this property. |
| W05-1529 32 9:45 2 SCF Acquisition System Following the design proposed by Briscoe and Carroll (1997), we built an SCF acquisition system consisting of the following four components: Charniaks parser (Charniak, 2000); an SCF extractor; a lemmatizer; and an SCF evaluator. |
| W07-1001 33 121:186 4.3 Parsing For automatic parsing, we made use of the wellknownCharniakparser(Charniak,2000). |
| H05-1102 34 196:366 5 Discussion and Future Work The parser proposed in this paper is an incremental parser, so the accuracy on dependency is lower than that for chart parsers, for example like those reported in (Collins, 1999; Charniak, 2000). |
| N03-1031 35 6:194 1 Introduction Current state-of-the-art statistical parsers (Collins, 1999; Charniak, 2000) are trained on large annotated corpora such as the Penn Treebank (Marcus et al. , 1993). |
| W05-1506 36 42:254 This generalization is not only of theoretical importance, but also critical in the application to state-of-theart parsers such as (Collins, 2003) and (Charniak, 2000). |
| W05-1506 37 39:254 They apply this method to the Charniak (2000) parser to get 50-best lists for reranking, yielding an improvement in parsing accuracy. |
| W05-1506 38 220:254 Collins (2000), in his parse-reranking experiments, used his Model 2 parser (Collins, 2003) with a beam width of 103 together with a cell limit of 100 to obtain k-best lists; the average number of parses obtained per sentence was 29.2, the maximum, 101.7 Charniak and Johnson (2005) use coarse-tone parsing on top of the Charniak (2000) parser and get 50-best lists for section 23. |
| W05-1506 39 216:254 We demonstrate this by comparing our k-best lists to those in (Ratnaparkhi, 1997), (Collins, 2000) and the parallel work by Charniak and Johnson (2005) in several ways, including oracle reranking and average number of found parses. |
| W05-1506 40 36:254 Since the original design of the algorithm described below, we have become aware of two efforts that are very closely related to ours, one by Jimenez and Marzal (2000) and another done in parallel to ours by Charniak and Johnson (2005). |
| W05-1506 41 191:254 It does support k-best parsing, but, following Collins parse-reranking work (Collins, 2000) (see also Section 5.1.2), it accomplishes this by simply abandoning dynamic programming, i.e., no items are considered equivalent (Charniak and Johnson, 2005). |
| W05-1506 42 235:254 60 86 88 90 92 94 96 98 1 2 5 10 20 30 50 70 100 Oracle F-score k (Charniak and Johnson, 2005) This work with beam width 10-4(Collins, 2000) (Ratnaparkhi, 1997) (a) Oracle Reranking 0 2 4 6 8 10 1 2 5 10 20 30 50 70 100 Percentage of Improvement over 1-best k (Charniak and Johnson, 2005) This work with beam width 10-4(Collins, 2000) (Ratnaparkhi, 1997) (b) Relative Improvement Figure 9: Absolutive and Relative F-scores of oracle reranking for the top k ( 100) parses for section 23, compared to (Charniak and Johnson, 2005), (Collins, 2000) and (Ratnaparkhi, 1997). |
| P02-1042 43 5:157 1 Introduction Most recent wide-coverage statistical parsers have used models based on lexical dependencies (e.g. Collins (1999), Charniak (2000)). |
| C02-1138 44 77:196 The other kind of treebank is the BLLIP corpus (Charniak, 2000). |
| N07-1053 45 85:200 parse-based features, we parse the TEXT elements of our documents with the Charniak parser (Charniak, 2000). |
| P04-1013 46 161:185 For comparison to previous results, table 2 lists the results for our best model (DGSSNFreq 20, rerank)9 and several other statistical parsers (Ratnaparkhi, 1999; Collins, 1999; Collins and Du y, 2002; Charniak, 2000; Collins, 2000; Bod, 2003) on the entire testing set. |
| P04-1013 47 21:185 2 Two History-Based Probability Models As with many previous statistical parsers (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000), we use a history-based model of parsing. |
| N03-1014 48 140:204 The bottom panel of table 1 lists the results for the two lexicalized models (SSN-Freq a1 200 and SSN-Freq a1 20) and five recent statistical parsers (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000; Collins, 2000; Bod, 2001). |
| N03-1014 49 58:204 The standard way to handle this problem is to hand-craft a finite set of features which provides a sufficient summary of the history (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000). |
| N03-1014 50 12:204 Previous approaches have used a hand-crafted finite set of features to represent the parse history (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000). |
| N03-1014 51 185:204 8 Related Work Most previous work on statistical parsing has used a history-based probability model with a hand-crafted set of features to represent the derivation history (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000). |
| N03-1014 52 186:204 Ratnaparkhi (1999) defines a very general set of features for the histories of a shift-reduce parsing model, but the results are not as good as models which use a more linguistically informed set of features for a top-down parsing model (Collins, 1999; Charniak, 2000). |
| N03-1014 53 169:204 The concept of lexical head is central to theories of syntax, and has often been used in designing hand-crafted history features (Collins, 1999; Charniak, 2000). |
| N03-1014 54 8:204 Many statistical parsers (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000) are based on a history-based model of parser actions. |
| W06-3806 55 76:86 We also used the following resources: the Charniak parser (Charniak, 2000) to carry out the syntactic analysis; the wn::similaritypackage (Pedersen et al. , 2004) to compute the Jiang&Conrath (J&C) distance (Jiang and Conrath, 1997) needed to implement the lexical similarity siml(T,H) as defined in (Corley and Mihalcea, 2005); SVM-lightTK (Moschitti, 2004) to encode the basic tree kernel function, KT, in SVM-light (Joachims, 1999). |
| P06-2041 56 160:166 For the English data, the result for SVM with model 5 is about 3 percentage points below the results obtained with the parser of Charniak (2000) and reported by Yamada and Matsumoto (2003). |
| P06-2041 57 6:166 1 Introduction Mainstream approaches in statistical parsing are based on nondeterministic parsing techniques, usually employing some kind of dynamic programming, in combination with generative probabilistic models that provide an n-best ranking of the set of candidate analyses derived by the parser (Collins, 1997; Collins, 1999; Charniak, 2000). |
| P02-1025 58 43:161 Strict left-to-right parsing makes it easy to combine with a standard 3-gram at the word level (Chelba and Jelinek, 2000) (Roark, 2001) rather than at sentence level (Charniak, 2001). |
| P02-1025 59 42:161 a74 previous results (Chelba and Jelinek, 2000) (Charniak, 2001) (Roark, 2001) show that a grammar-based language model benefits from interpolation with a 3-gram model. |
| P02-1025 60 70:161 Since the SLM parses sentences bottom-up while the parsers used in (Charniak, 2000), (Charniak, 2001) and (Roark, 2001) are top-down, its not clear how to find a direct correspondence between our schemes of enriching the dependency structure and the ones employed above. |
| P02-1025 61 9:161 The statistical parsing community has used various ways of enriching the dependency structure underlying the parametrization of the probabilistic model used for scoring a given parse tree (Charniak, 2000) (Collins, 1999). |
| P04-1040 62 31:199 In the experiments we used the parsers described in (Charniak, 2000) and (Collins, 1999). |
| P04-1040 63 27:199 Blaheta and Charniak (2000) presented the first method for assigning Penn functional tags to constituents identified by a parser. |
| P04-1040 64 15:199 The evaluation of the transformed output of the parsers of Charniak (2000) and Collins (1999) gives 90% unlabelled and 84% labelled accuracy with respect to dependencies, when measured against a dependency corpus derived from the Penn Treebank. |
| P04-1040 65 110:199 First, because of the different definition of a correctly identified constituent in the parsers output, we apply our method to a greater portion of all labels produced by the parser (95% vs. 89% reported in (Blaheta and Charniak, 2000)). |
| P04-1040 66 184:199 TiMBL performed well on tasks where structured, more complicated and task-specific statistical models have been used previously (Blaheta and Charniak, 2000). |
| P04-1040 67 101:199 Thus, it is informative to compare our results with those reported in (Blaheta and Charniak, 2000) for this same task. |
| P04-1040 68 79:199 As an alternative to hardcoded heuristics, Blaheta and Charniak (2000) proposed to recover the Penn functional tags automatically. |
| P05-1024 69 150:200 CH00 = (Charniak, 2000), CO00=(Collins, 2000). |
| P05-1024 70 139:200 (Charniak, 2000) extends PCFG and achieves similar performance to (Collins, 2000). |
| P02-1047 71 24:159 To build such systems, we train a family of Naive Bayes classifiers on a large set of examples that are generated automatically from two corpora: a corpus of 41,147,805 English sentences that have no annotations, and BLIPP, a corpus of 1,796,386 automatically parsed English sentences (Charniak, 2000), which is available from the Linguistic Data Consortium (www.ldc.upenn.edu). |
| P02-1047 72 102:159 To each text span in the BLIPP corpus corresponds a parse tree (Charniak, 2000). |
| P02-1047 73 61:159 The second, called BLIPP, is a corpus of only 1,796,386 sentences that were parsed automatically by Charniak (2000). |
| W06-3119 74 20:125 We then use Charniaks parser (Charniak, 2000) to generate the most likely parse tree for each English target sentence in the training corpus. |
| W07-2218 75 18:279 (Charniak, 2000; Collins, 1999; Nivre et al. , 2004)). |
| W07-2218 76 9:279 All the most accurate dependency parsing models are fully discriminative, unlike constituent parsing where all the state of the art methods have a generative component (Charniak and Johnson, 2005; Henderson, 2004; Collins, 2000). |
| D07-1117 77 116:217 First a bracketing parser (the Charniak parser (Charniak, 2000) in our case) is used to generate the parse tree of a sentence, then the const2dep tool developed by Hwa was utilized to convert the bracketing tree to a dependency tree based on the head percolation table developed by the second author. |
| W05-1514 78 178:207 LR LP F-score Ratnaparkhi (1997) 86.3 87.5 86.9 Collins (1999) 88.1 88.3 88.2 Charniak (2000) 89.6 89.5 89.5 Kudo (2005) 89.3 89.6 89.4 Sang (2001) 78.7 82.3 80.5 Deterministic (tagger-POSs) 81.2 86.5 83.8 Deterministic (gold-POSs) 82.6 87.7 85.1 Search (tagger-POSs) 83.2 87.1 85.1 Search (gold-POSs) 84.6 88.5 86.5 Iterative Search (tagger-POSs) 85.0 86.8 85.9 Iterative Search (gold-POSs) 86.2 88.0 87.1 Table 6: Comparison with other work. |
| W05-1514 79 13:207 However, there is a very large gap between their performance and that of widely-used practical parsers (Charniak, 2000; Collins, 1999). |
| N06-1024 80 76:144 7And the (Charniak, 2000) parser that (Blaheta, 2003) used has a reported F-measure of 89.5, higher than the Bikel parser used here. |
| N06-1024 81 122:144 Note that our systems parsed scores were obtained using our modified version of Bikels implementation of Collinss thesis parser which assigns function tags, while the other PSLB postprocessing systems use Charniaks parser (Charniak, 2000) and Dienes and Dubey integrate empty category recovery directly into a variant of Collinss parser. |
| N06-1024 82 67:144 For purposes of comparison, we have calculated our overall score both with and without CLR.5 The (Blaheta, 2003) numbers in parentheses in Table 1 are from hisfeaturetreesspecializedfortheSyntacticandSemantic groups, while all his other numbers, including the overall score, are from using a single feature set for his four function tag groups.6 5(Jijkoun and de Rijke, 2004) do not state whether they are including CLR, but since they are comparing their results to (Blaheta and Charniak, 2000), we are assuming that they do. |
| N06-1024 83 8:144 Modern statistical parsers such as (Collins, 2003) and (Charniak, 2000) however ignore much of this information and return only an We would like to thank Fernando Pereira, Dan Bikel, Tony Kroch and Mark Liberman for helpful suggestions. |
| P03-1055 84 7:180 However, such constructions prove to be difficult for stochastic parsers (Collins et al. , 1999) and they either avoid tackling the problem (Charniak, 2000; Bod, 2003) or only deal with a subset of the problematic cases (Collins, 1997). |
| P01-1010 85 50:160 4.1 The base line For our base line parse accuracy, we used the now standard division of the WSJ (see Collins 1997, 1999; Charniak 1997, 2000; Ratnaparkhi 1999) with sections 2 through 21 for training (approx. |
| P01-1010 86 69:160 While most subtree restrictions diminish the accuracy scores, we will see that there are restrictions that improve our scores, even beyond those of Charniak (2000). |
| P01-1010 87 123:160 These scores slightly outperform the best previously published parser by Charniak (2000), who obtained 89.5% LP and 89.6% LR for test sentences 100 words. |
| P01-1010 88 105:160 Head-lexicalized stochastic grammars have recently become increasingly popular (see Collins 1997, 1999; Charniak 1997, 2000). |
| P01-1010 89 66:160 Table 1 shows the LP and LR scores obtained with our base line subtree set, and compares these scores with those of previous stochastic parsers tested on the WSJ (respectively Charniak 1997, Collins 1999, Ratnaparkhi 1999, and Charniak 2000). |
| P01-1010 90 67:160 The table shows that by using the base line subtree set, our parser outperforms most previous parsers but it performs worse than the parser in Charniak (2000). |
| P01-1010 91 158:160 The importance of including single nonheadwords is now also uncontroversial (e.g. Collins 1997, 1999; Charniak 2000), and the current paper has shown the importance of including two and more nonheadwords. |
| P01-1010 92 141:160 Charniak 1996, 1997), while most current stochastic parsing models use a "markov grammar" (e.g. Collins 1999; Charniak 2000). |
| P01-1010 93 154:160 context-free rules Charniak (1996) Collins (1996), Eisner (1996) context-free rules, headwords Charniak (1997) context-free rules, headwords, grandparent nodes Collins (2000) context-free rules, headwords, grandparent nodes/rules, bigrams, two-level rules, two-level bigrams, nonheadwords Bod (1992) all fragments within parse trees Scope of Statistical Dependencies Model Figure 4. |
| P01-1010 94 87:160 Our highest scores of 90.8% LP and 90.5% LR outperform the scores of the best previously published parser by Charniak (2000) who obtains 90.1% for both LP and LR. |
| P01-1010 95 80:160 The highest scores are obtained if the full base line subtree set is used, but they remain behind the results of Charniak (2000). |
| P04-1082 96 80:164 5 Evaluation Following Johnson (2002), the system was evaluated on two different kinds of input: first, on perfect input, i.e., PTB annotations stripped of all empty categories and information related to them; and second, on imperfect input, in this case the output of Charniaks (2000) parser. |
| P04-1082 97 62:164 Charniaks parser (Charniak, 2000), however, does not include function tags, so in order for the algorithm to work properly on parser output (see Section 5), additional functions were written to approximate the required tags. |
| P04-1082 98 12:164 State-of-the-art statistical parsers (e.g. Charniak, 2000) typically provide a labeled bracketing only; i.e., a parse tree without empty categories. |
| P04-1082 99 63:164 Presumably, the accuracy of the algorithm on parser output would be enhanced by accurate prior assignment of the tags to all relevant nodes, as in Blaheta and Charniak (2000) (see also Section 5). |
| P04-1082 100 101:164 As mentioned in Section 4, it is believed that the results of the current method on parser output would improve if that output were reliably assigned function tags, perhaps along the lines of Blaheta and Charniak (2000). |
| P04-1082 101 93:164 5.2 Parser output The system was also run using as input the output of Charniaks parser (Charniak, 2000). |
| P04-1082 102 6:164 1 Introduction Many recent approaches to parsing (e.g. Charniak, 2000) have focused on labeled bracketing of the input string, ignoring aspects of structure that are not reflected in the string, such as phonetically null elements and long-distance dependencies, many of which provide important semantic information such as predicate-argument structure. |
| N01-1029 103 117:249 The second set was trained on the BLLIP WSJ Corpus (BWC), which is a machine-parsed (Charniak, 2000) version of (a selection of) the ACL/DCI corpus, very similar to the selection made for the WSJ0/1 CSR corpus. |
| W05-1509 104 28:202 Following (Blaheta and Charniak, 2000), we concentrate on syntactic and semantic function labels. |
| W05-1509 105 155:202 Following (Blaheta and Charniak, 2000), incorrectly parsed constituents will be ignored (roughly 11% of the total) in the evaluation of the precision and recall of the function labels, but not in the evaluation of the parser. |
| W05-1509 106 11:202 duce trees annotated with bare phrase structure labels (Collins, 1999; Charniak, 2000). |
| W05-1509 107 165:202 Previous work that uses, like us, a single model for both types of labels reaches an F measure of 95.7% for syntactic labels and 79.0% for semantic labels (Blaheta and Charniak, 2000). |
| W05-1509 108 153:202 Individual performance on syntactic and semantic function labelling compare favourably to previous attempts (Blaheta, 2004; Blaheta and Charniak, 2000). |
| W05-1509 109 52:202 As with many other statistical parsers (Collins, 1999; Charniak, 2000), SSN parsers use a history-based model of parsing. |
| W05-1509 110 14:202 Unlike phrase structure labels, function labels are contextdependent and encode a shallow level of phrasal and lexical semantics, as observed first in (Blaheta and Charniak, 2000). |
| W05-1509 111 25:202 Specifically, the parser outputs additional labels indicating the function of a constituent in the tree, such as NP-SBJ or PP-TMP in the tree 1(Blaheta and Charniak, 2000) talk of function tags.We will instead use the term function label, to indicate function identifiers, as they can decorate any node in the tree. |
| P05-1013 112 130:139 Although the best published results for the Collins parser is 80% UAS (Collins, 1999), this parser reaches 82% when trained on the entire training data set, and an adapted version of Charniaks parser (Charniak, 2000) performs at 84% (Jan Hajic, pers. |
| W07-0605 113 19:174 (Charniak, 2000; Briscoe et al. , 2006), have wide coverage and yield grammatical representations capable of supporting various applications (e.g. summarization, information extraction). |
| P05-1072 114 21:208 We investigate ways to combine hypotheses generated from semantic role taggers trained using different syntactic views one trained using the Charniak parser (Charniak, 2000), another on a rule-based dependency parser Minipar (Lin, 1998), and a third based on a flat, shallow syntactic chunk representation (Hacioglu, 2004a). |
| J05-1003 115 65:603 First, several of the best-performing parsers on the WSJ treebank (e.g. , Ratnaparkhi 1997; Charniak 1997, 2000; Collins 1997, 1999; Henderson 2003) are cases of history-based models. |
| J05-1003 116 505:603 The model in Charniak (2000) is quite different, however. |
| J05-1003 117 504:603 Our features are in many ways similar to those of Charniak (2000). |
| J05-1003 118 439:603 The method gives very similar accuracy to the model of Charniak (2000), which also uses a rich set of initial features in addition to Charniaks (1997) original model. |
| J05-1003 119 502:603 Related Work 6.1 History-Based Models with Complex Features Charniak (2000) describes a parser which incorporates additional features into a previously developed parser, that of Charniak (1997). |
| W03-1005 120 5:166 1 Introduction Many broad-coverage statistical parsers (Charniak, 2000; Collins, 1999; Bod, 2001) are not able to give a full interpretation for sentences such as: (1) It is difficult to guess what she wants to buy. |
| W08-2102 121 20:209 For example, the model in (Taskar et al., 2004) is trained on only sentences of 15 words or less; reranking models (Collins, 2000; Charniak and Johnson, 2005) restrict Y(x) to be a small set of parses from a first-pass parser; see section 1.1 for discussion of other related work. |
| W08-2102 122 31:209 Experiments on the Penn WSJ treebank show that the model recovers constituent structures with higher accuracy than the approaches of (Charniak, 2000; Collins, 2000; Petrov and Klein, 2007), and with a similar level of performance to the reranking parser of (Charniak and Johnson, 2005). |
| W08-2102 123 35:209 In reranking approaches, a first-pass parser is used to enumerate a small set of candidate parses for an input sentence; the reranking model, which is a GLM, is used to select between these parses (e.g., (Ratnaparkhi et al., 1994; Johnson et al., 1999; Collins, 2000; Charniak and Johnson, 2005)). |
| W08-2102 124 183:209 Our experiments show an improvement in performance over the results in (Collins, 2000; Charniak, 2000). |
| W08-2102 125 185:209 The Charniak (2000) model is also arguably more complex, again using a carefully constructed generative model. |
| W08-2102 126 159:209 (2008), Charniak (2000), Collins (2000), Petrov and Klein (2007), Charniak and Johnson (2005), and Huang (2008). |
| W08-2102 127 189:209 Charniak and Johnson (2005), and Huang (2008), describe approaches that make use of nonlocal features in conjunction with the Charniak (2000) model; future work may consider extending our approach to include non-local features. |
| W07-0604 128 130:187 (2005) report 86.9% LAS on about 2,000 words of Eve data, using the Charniak (2000) parser with a separate dependency-labeling step. |
| W07-0604 129 177:187 That work relied on a phrase-structure statistical parser (Charniak, 2000) trained on the Penn Treebank, and the output of that parser had to be converted into CHILDES grammatical relations. |
| W07-0604 130 90:187 (2004) has been used in practice for automatic parsing of child language transcripts (Sagae et al. , 2004; Sagae et al. , 2005), such work relied mainly on a statistical parser (Charniak, 2000) trained on the Wall Street Journal portion of the Penn Treebank, since a large enough corpus of annotated CHILDES data was not available to train a domain-specific parser. |
| D07-1062 131 47:201 For the test data, we report on results using the gold-standardTreebankdata,andinadditionwealso report results on automatically parsed data using the Charniak parser (Charniak, 2000) as provided by the CoNLL 2005 shared task. |
| D07-1062 132 128:201 Tofullyevaluatetheinfluenceofthe LTAG-basedfeatures,wereporttheidentificationresults on both Gold Standard parses and on Charniak parser output (Charniak, 2000)5. |
| P06-2048 133 5:249 1 Introduction Much of the current research into probabilistic parsing is founded on probabilistic contextfree grammars (PCFGs) (Collins, 1996; Charniak, 1997; Collins, 1999; Charniak, 2000; Charniak, 2001; Klein and Manning, 2003). |
| H05-1099 134 133:185 Note that the systems used here are exactly the ones presented for the original Li & Roth task, in SecPunctuation System Leave Ignore Li & Roth (reference tags) 88.47 SPRep avg perceptron Reference tags 91.37 91.86 Brill tags 87.94 88.42 Charniak (2000) 87.94 88.44 Unweighted intersection 88.66 89.16 Weighted intersection 89.22 89.69 Table 4: Shallow bracketing accuracy of several different systems, trained on sections 2-21 of WSJ Treebank and applied to section 4 of the Switchboard Treebank. |
| H05-1099 135 97:185 3 Experimental Results 3.1 Comparing Finite-State and Context-Free Parsers The first two rows of Table 3 present a comparison between the SPRep shallow parser and the Charniak (2000) context-free parser detailed in Charniak and Johnson (2005). |
| H05-1099 136 70:185 2.2 Shallow Parser In addition to the trainable n-best context-free parser from Charniak (2000), we needed a trainable shallow parser to apply to the variety of tasks we were interested in investigating. |
| H05-1099 137 68:185 mentation of the Collins parser and the n-best version of the Charniak (2000) parser, documented in Charniak and Johnson (2005), fit the requirements. |
| H05-1099 138 62:185 Of the two parsers we evaluated, the Charniak (2000) parser gave the best performance, which is consistent with its higher reported performance on the context-free parsing task versus other context-free parsers. |
| H05-1099 139 78:185 The one-best performance of the parser is the same as what was presented in Charniak (2000). |
| H05-1099 140 11:185 The output of a contextfree parser, such as that of Collins (1997) or Charniak (2000), can be transformed into a sequence of shallow constituents for comparison with the output of a shallow parser. |
| H05-1099 141 120:185 Again, the best-scoring candidate using this composite score is selected from among the shallow 791 NP-Chunking CoNLL-2000 Li & Roth task Punctuation Punctuation Punctuation System Leave Ignore Leave Ignore Leave Ignore SPRep averaged perceptron 94.21 94.25 93.54 93.70 95.12 95.27 Charniak (2000) 94.17 94.20 93.77 93.92 95.15 95.32 Unweighted intersection 95.13 95.16 94.52 94.64 95.77 95.92 Weighted intersection 95.57 95.58 95.03 95.16 96.20 96.33 Table 3: F-measure shallow bracketing accuracy on three shallow parsing tasks, for the SPRep perceptron shallow parser, the Charniak (2000) context-free parser, and for systems combining the SPRep and Charniak system outputs. |
| H05-1099 142 100:185 For these trials, we used just the one-best output of that model, which is the same as in Charniak (2000). |
| H05-1099 143 63:185 Collins (2000) reported a reranking model that improved his parser output to roughly the same level of accuracy as Charniak (2000), and Charniak and Johnson (2005) report an improvement using reranking over Charniak (2000). |
| H05-1099 144 151:185 The Charniak and Johnson (2005) system output (denoted C & J in the table) before reranking (denoted one-best) is identical to the Charniak (2000) results that have been reported in the other tables. |
| H05-1099 145 42:185 Perhaps the most widely accepted convention is that of ignoring punctuation for the purposes of assigning constituent span, under the perspective that, fun788 Phrase Evaluation Scenario System Type (a) (b) (c) Modified All 98.37 99.72 99.72 Truth VP 92.14 98.70 98.70 Li and Roth All 94.64 (2001) VP 95.28 Collins (1997) All 92.16 93.42 94.28 VP 88.15 94.31 94.42 Charniak All 93.88 95.15 95.32 (2000) VP 88.92 95.11 95.19 Table 1: F-measure shallow bracketing accuracy under three different evaluation scenarios: (a) baseline, used in Li and Roth (2001), with original chunklink script converting treebank trees and context-free parser output; (b) same as (a), except that empty subject NPs are inserted into every unary SVP production; and (c) same as (b), except that punctuation is ignored for setting constituent span. |
| P06-1041 146 6:175 Stochastic parsers for English trained on the Penn Treebank have peaked their performance around 90% (Charniak, 2000). |
| W04-0834 147 50:87 3.4 Syntactic Relations We first parse the sentence containinga0 with a statistical parser (Charniak, 2000). |
| W08-0909 148 36:236 Rather than using sentence lengthasa proxy, measurescanemploy toolsforautomatic analysis of the syntactic structure of texts (e.g., (Charniak, 2000)). |
| H05-1033 149 16:234 Another important reason is the development of robust syntactic parsers (e.g. , Charniak, 2000) that can be used to provide critical structural and lexical information to the discourse parser. |
| H05-1033 150 210:234 Both the baseline and Spade operate on parse trees which were obtained from Charniaks (2000) parser. |
| W05-1516 151 140:171 Performance of Alternative Models 157 5 Related Work Previous parsing models (e.g. , Collins, 1997; Charniak, 2000) maximize the joint probability P(S, T) of a sentence S and its parse tree T. We maximize the conditional probability P(T | S). |
| P06-3009 152 49:147 Parsing models have been developed for different languages and state-of-the-art results have been reported for, e.g., English (Collins, 1997; Charniak, 2000). |
| W05-1518 153 46:283 English version described in Charniak (2000), Czech adaptation 2002 2003, unpublished. |
| W06-2904 154 8:160 1 Introduction Over the past decade, there has been tremendous progress on learning parsing models from treebank data (Collins, 1997; Charniak, 2000; Wang et al. , 2005; McDonald et al. , 2005). |
| W06-2904 155 11:160 Subsequent research began to focus more on conditional models of parse structure given the input sentence, which allowed discriminative training techniques such as maximum conditional likelihood (i.e. maximum entropy ) to be applied (Ratnaparkhi, 1999; Charniak, 2000). |
| P08-1006 156 7:181 1 Introduction Parsing technologies have improved considerably in the past few years, and high-performance syntactic parsers are no longer limited to PCFG-based frameworks (Charniak, 2000; Klein and Manning, 2003; Charniak and Johnson, 2005; Petrov and Klein, 2007), but also include dependency parsers (McDonald and Pereira, 2006; Nivre and Nilsson, 2005; Sagae and Tsujii, 2007) and deep parsers (Kaplan et al., 2004; Clark and Curran, 2004; Miyao and Tsujii, 2008). |
| P08-1006 157 35:181 NO-RERANK Charniak (2000)s parser, based on a lexicalized PCFG model of phrase structure trees.3 The probabilities of CFG rules are parameterized on carefully hand-tuned extensive information such as lexical heads and symbols of ancestor/sibling nodes. |
| W05-0211 158 74:173 In (Charniak, 2000), he proposes a generative model based on a Markov-grammar. |
| P08-2054 159 7:74 Treebank-specificheuristicshavecommonlybeen used both to alleviate inadequate independence assumptions stipulated by naive PCFGs (Collins, 1999; Charniak, 2000). |
| C08-1027 160 105:249 The English side is parsed using a state-of-the-art statistical English parser (Charniak, 2000). |
| D08-1052 161 20:235 The designs of hypotheses in (Collins, 1999; Charniak, 2000) show a delicate balance between expressiveness and tractability, which play an important role in natural language parsing. |
| W07-2028 162 17:85 A tree kernel (Moschitti, 2004) is used to exploit the deep syntactic processing obtained using the Charniak parser (Charniak, 2000). |
| C00-2135 163 15:148 The progress in parsing technology are noteworthy, and in particular, various statistical dependency models have been proposed(Collins, 1997),, (Ratnaparkhi, 1997), (Charniak, 2000). |
| P04-2009 164 83:110 5.1 Parsers used Charniaks parser (2000) is a combination probabilistic context free grammar and maximum entropy parser. |
| W08-0401 165 140:232 For extraction of source sentence tree structure, we used the Charniak parser (Charniak 2000). |
| J06-2003 166 514:682 In order to use this tool, we parsed the English sentences with Charniaks parser (Charniak 2000). |
| H05-2004 167 21:45 Next, sentences are analyzed by a state-of-the-art syntactic parser (Charniak, 2000) the output of which provides useful information for the main SRL module. |
| N06-1023 168 28:177 Recently, many accurate statistical parsers have been proposed (e.g. , (Collins, 1999; Charniak, 2000) for English, (Uchimoto et al. , 2000; Kudo and Matsumoto, 2002) for Japanese). |
| N06-1023 169 11:177 Blaheta and Charniak proposed a statistical method Currently, National Institute of Information and Communications Technology, JAPAN, dk@nict.go.jp Currently, Graduate School of Informatics, Kyoto University, kuro@i.kyoto-u.ac.jp for analyzing function tags in Penn Treebank, and achieved a really high accuracy of 95.7% for syntactic roles, such as SBJ (subject) and DTV (dative) (Blaheta and Charniak, 2000). |
| J03-4003 170 682:797 Finally, Bod (2001) describes a very different approach (a DOP approach to parsing) that gives excellent results on treebank parsing, comparable to the results of Charniak (2000) and Collins (2000). |
| J03-4003 171 675:797 Charniak (2000) describes a parsing model that also uses Markov processes to generate rules. |
| J03-4003 172 667:797 Blaheta and Charniak (2000) describe methods for the recovery of the semantic tags in the Penn Treebank annotations, a significant step forward from the complement/adjunct distinction recovered in model 2 of the current article. |
| J03-4003 173 343:797 Two models (Collins 2000; Charniak 2000) outperform models 2 and 3 on section 23 of the treebank. |
| J03-4003 174 345:797 Charniak (2000) describes a series of enhancements to the earlier model of Charniak (1997). |
| J03-4003 175 713:797 Charniak (2000) shows that using the POS tags as word class information in the model is important for parsing accuracy. |
| P06-3014 176 7:139 1 Introduction Robust statistical syntactic parsers, made possible by new statistical techniques (Collins, 1999; Charniak, 2000; Bikel, 2004) and by the availability of large, hand-annotated training corpora such as WSJ (Marcus et al. , 1993) and Switchboard (Godefrey et al. , 1992), have had a major impact on the field of natural language processing. |
| P06-3014 177 126:139 All these results seem to suggest that adding punctuation in speech transcription is of little help to statistical parsers including at least three state-of-the-art statistical parsers (Collins, 1999; Charniak, 2000; Bikel, 2004). |
| P07-1120 178 30:209 The Charniak parsing pipeline has been extensively studied over the past decade, with a number of papers focused on improving early stages of the pipeline (Charniak et al. , 1998; Caraballo and Charniak, 1998; Blaheta and Charniak, 1999; Hall and Johnson, 2004; Charniak et al. , 2006) as well as many focused on optimizing final parse accuracy (Charniak, 2000; Charniak and Johnson, 2005; McClosky et al. , 2006). |
| P07-1120 179 33:209 The well-known Charniak (2000) coarse-to-fine parser is a two-stage parsing pipeline, in which the first stage uses a vanilla PCFG to populate a chart of parse constituents. |
| P07-1120 180 9:209 Pipeline systems are ubiquitous in natural language processing, used not only in parsing (Ratnaparkhi, 1999; Charniak, 2000), but also machine translation(OchandNey, 2003)andspeechrecognition (Fiscus, 1997; Goel et al. , 2000), among others. |
| N06-1020 181 51:208 3.1 The first-stage 50-best parser The first stage of our parser is the lexicalized probabilistic context-free parser described in (Charniak, 2000) and (Charniak and Johnson, 2005). |
| N06-1020 182 8:208 Given sufficient labelled data, there are several supervised techniques of training high-performance parsers (Charniak and Johnson, 2005; Collins, 2000; Henderson, 2004). |
| D07-1078 183 19:178 Binarizing the syntax trees for syntax-based machine translation is similar in spirit to generalizing parsing models via markovization (Collins, 1997; Charniak, 2000). |
| J06-1005 184 281:684 Each sentence in this corpus was then parsed using the syntactic parser developed by Charniak (2000). |
| J06-1005 185 114:684 7 Besides the work on semantic roles (Charniak 2000; Gildea and Jurafsky 2002; Thompson, Levy, and Manning 2003), considerable interest has been shown in the automatic interpretation of various noun phrase-level constructions, such as noun compounds. |
| W03-1017 186 67:165 Syntactic structure was obtained with Charniaks statistical parser (Charniak, 2000). |
| W07-1209 187 67:185 This parser uses a discriminative reranker that selects the most probable parse from the 50-best parses returned by a generative parser based on Charniak (2000). |
| W05-1513 188 133:152 Table 1 shows a summary of the results of our experiments with SVMpar and MBLpar, and also results obtained with the Charniak (2000) parser, the Bikel (2003) implementation of the Collins (1997) parser, and the Ratnaparkhi (1997) parser. |
| W05-1513 189 10:152 Although state-of-the-art statistical parsers (Collins, 1997; Charniak, 2000) are more accurate, the simplicity and efficiency of deterministic parsers make them attractive in a number of situations requiring fast, light-weight parsing, or parsing of large amounts of data. |
| W02-1039 190 43:151 When an S alignment exists, there will always also exist a P alignment such that P a65 S. The English sentences were parsed using a state-of-the-art statistical parser (Charniak, 2000) trained on the University of Pennsylvania Treebank (Marcus et al. , 1993). |
| P07-1062 191 65:192 Secondly, Charniak and Johnson (2005) showed how reranking of the 50best output of the Charniak (2000) parser gives substantial improvements in parsing accuracy. |
| P07-1062 192 14:192 Their models and algorithm subsequently packaged together into the publicly available SPADE discourse parser1 make use of the output of the Charniak (2000) parser to derive syntactic indicator features for segmentation and discourse parsing. |
| W07-0208 193 26:151 Here, solid lines correspond to surface syntactic structure, produced by Charniaks parser (Charniak, 2000), and dashed lines are an encoding of the Proposition Bank annotation of the semantic roles with respect to the verb stopped. |
| W07-0208 194 30:151 Consider the task of recovering non-local dependencies (such as control, WH-extraction, topicalization) in the surface syntactic phrase trees produced by the state-of-the-art parser of (Charniak, 2000). |
| N03-1011 195 91:172 Each sentence in this corpus was then parsed using the syntactic parser developed by Charniak (Charniak, 2000). |
| W02-1006 196 66:163 1(a) attention (noun) 1(b) He turned his attention to the workbench . 1(c) a24 turned, VBD, active, lefta41 2(a) turned (verb) 2(b) He turned his attention to the workbench . 2(c) a24 he, attention, PRP, NN, VBD, activea41 3(a) green (adj) 3(b) The modern tram is a green machine . 3(c) a24 machine, NNa41 Table 1: Examples of syntactic relations (assuming no feature selection) 3.4 Syntactic Relations We first parse the sentence containing a2 with a statistical parser (Charniak, 2000). |
| H05-1078 197 64:199 Following (Blaheta and Charniak, 2000), we refer to the first class as syntactic function labels, and to the second class as semantic function labels. |
| H05-1078 198 47:199 As with many other statistical parsers (Collins, 1999; Charniak, 2000), the model of parsing is history-based. |
| H05-1078 199 30:199 While the function of a constituent and its structural position are often correlated, they some1(Blaheta and Charniak, 2000) talk of function tags. |
| H05-1078 200 181:199 First, they parse the Penn Treebank using a state-of-the-art parser (Charniak, 2000). |
| H05-1078 201 179:199 In work that predates the availability of Framenet and Propbank, (Blaheta and Charniak, 2000) define the task of function labelling for the first time and highlight its relevance for NLP. |
| H05-1078 202 171:199 We will therefore discuss separately those pieces of work that have made limited use of function labels for parsing (Klein and Manning, 2003), and those that have concentrated on recovering function labels as a separate task (Blaheta and Charniak, 2000; Blaheta, 2004). |
| H05-1078 203 22:199 tion labels are context-dependent and encode a shallow level of phrasal and lexical semantics, as observed first in (Blaheta and Charniak, 2000).1 To a large extent, they overlap with semantic role labels as defined in PropBank (Palmer et al. , 2005). |
| H05-1078 204 17:199 Statistical parsers trained on the Penn Treebank (PTB) (Marcus et al. , 1993) produce trees annotated with bare phrase structure labels (Collins, 1999; Charniak, 2000). |
| H05-1078 205 66:199 Like previous work (Blaheta and Charniak, 2000), we complete the sets of syntactic and semantic labels by labelling constituents that do not bear any function label with a NULL label.3 2.3 Evaluation To evaluate the performance of our function parsing experiments, we will use several measures. |
| H05-1078 206 79:199 Following (Blaheta and Charniak, 2000), incorrectly parsed constituents will be ignored (roughly 11% of the total) in the evaluation of the precision and recall of the function labels, but not in the evaluation of the parser. |
| W01-0714 207 7:226 Because of these kinds of results, the vast majority of statistical parsing work has focused on parsing as a supervised learning problem (Collins, 1997; Charniak, 2000). |
| P05-1073 208 15:185 We present performance results on the February 2004 version of PropBank on gold-standard parse trees as well as results on automatic parses generated by Charniaks parser (Charniak, 2000). |
| P05-1073 209 165:185 We trained and tested on automatic parse trees from Charniaks parser (Charniak, 2000). |
| W04-0307 210 96:154 As in (Ratnaparkhi, 1999; Charniak, 2000; Collins, 1999), we evaluate on all sentences with length 40 words (2,245 sentences) and length 100 words (2,416 sentences). |
| W04-0307 211 5:154 1 Introduction Statistical parsing has been an important focus of recent research (Magerman, 1995; Eisner, 1996; Charniak, 1997; Collins, 1999; Ratnaparkhi, 1999; Charniak, 2000). |
| W04-0307 212 9:154 Charniak (Charniak, 2000) developed a state-of-the-art statistical CFG parser and then built an effective language model based on it (Charniak, 2001). |
| W04-0307 213 145:154 Models 40 words (2,245 sentences) Without TRACE All (1,903 sentences) (2,245 sentences) governor only all roles governor only all roles RLP RLR RLP RLR RLP RLR RLP RLR L 92.4 92.4 89.5 88.7 92.4 92.3 89.1 88.6 T 93.2 92.9 89.9 89.3 93.2 92.9 89.8 89.2 Charniak (Charniak, 2000) 92.6 92.5 89.4 88.9 92.5 92.3 88.9 88.7 Collins, Model 2 (Collins, 1999) 92.5 92.3 89.1 88.5 92.2 92.1 89.0 88.5 Collins, Model 3 (Collins, 1999) 92.8 92.7 89.9 89.4 92.7 92.4 89.3 89.1 Models 100 words (2,416 sentences) Without TRACE All (1,979 sentences) (2,416 sentences) governor only all roles governor only all roles RLP RLR RLP RLR RLP RLR RLP RLR L 91.9 91.6 88.8 88.1 91.8 91.5 88.5 87.8 T 92.7 92.3 89.4 88.7 92.6 92.2 89.1 88.5 Charniak (Charniak, 2000) 92.0 91.8 88.8 88.2 91.9 91.6 88.4 87.9 Collins, Model 2 (Collins, 1999) 91.8 91.6 88.6 88.0 91.7 91.5 88.2 87.9 Collins, Model 3 (Collins, 1999) 92.2 92.1 89.4 88.8 92.1 91.9 88.8 88.5 CFG parsers may loose accuracy from the CFG-toCDG transformation, similarly to Collins experiment reported in (Hajic et al. , 1998), we also transformed our CDG parses to Penn Treebank style CFG parse trees and scored them using PARSEVAL. |
| W04-0307 214 128:154 5.2 Comparing to Other Parsers Charniaks state-of-the-art PCFG parser (Charniak, 2000) has achieved the highest PARSEVAL LP/LR when compared to Collins Model 2 and Model 3 (Collins, 1999), Roarks (Roark, 2001), Ratnaparkhis (Ratnaparkhi, 1999), and Xu & Chelbas (Xu et al. , 2002) parsers. |
| W04-0307 215 6:154 Several of these parsers generate constituents by conditioning probabilities on non-terminal labels, part-of-speech (POS) tags, and some headword information (Collins, 1999; Ratnaparkhi, 1999; Charniak, 2000). |
| C08-1094 216 8:177 The Charniak (2000) parser uses a simple PCFG to prune the chart for a richer model; and Charniak and Johnson (2005) added a discriminatively trained reranker to the end of that pipeline. |
| C08-1094 217 97:177 5 Constraining the Charniak parser 5.1 Parser overview and constraint methods The Charniak (2000) parser is a multi-stage, agenda-driven, edge-based parser, that can be constrained by precluding edges from being placed on the agenda. |
| W04-2407 218 129:153 1999; Charniak, 2000). |
| W04-2407 219 86:153 Thus, the Penn Treebank of American English (Marcus et al. , 1993) has been used to train and evaluate the best available parsers of unrestricted English text (Collins, 1999; Charniak, 2000). |
| W07-2053 220 66:89 We use MXPOST tagger (Adwait, 1996) for POS tagging, Charniak parser (Charniak, 2000) for extracting syntactic relations, and David Blei?s version of LDA1 for LDA training and inference. |
| D07-1108 221 103:175 We use MXPOST tagger (Adwait, 1996) for POS tagging, Charniak parser (Charniak, 2000) for extracting syntactic relations, SVMlight1 for SVM classifier and David Bleis version of LDA2 for LDA training and inference. |
| W04-2002 222 49:60 This finding is in line with Charniaks own analysis, which shows that the high performance of his model is due to the fact that it combines a thirdorder Markov grammar with sophisticated phrasal and lexical features (Charniak, 2000). |
| W04-2002 223 39:60 These models include a standard unlexicalized PCFG parser, a head-lexicalized parser (Collins, 1997), and a maximum-entropy inspired parser (Charniak, 2000). |
| P02-1018 224 115:149 Then as is standard, the precision P, recall R and f-score f are calculated as follows: P = jG \ TjjTj R = jG \ TjjGj f = 2 P RP + R Table 3 provides these measures for two different test corpora: (i) a version of section 23 of the Penn Treebank from which empty nodes, indices and unary branching chains consisting of nodes of the same category were removed, and (ii) the trees produced by Charniaks parser on the strings of section 23 (Charniak, 2000). |
| P02-1018 225 4:149 Evaluating the algorithm on the output of Charniaks parser (Charniak, 2000) and the Penn treebank (Marcus et al. , 1993) shows that the patternmatching algorithm does surprisingly well on the most frequently occuring types of empty nodes given its simplicity. |
| P02-1018 226 8:149 Broad coverage syntactic parsers with good performance have recently become available (Charniak, 2000; Collins, 2000), but these typically produce as output a parse tree that only encodes local syntactic information, i.e., a tree that does not include any empty nodes. |
| P02-1018 227 129:149 The parser output was produced by Charniaks parser (Charniak, 2000). |
| W06-2903 228 7:150 Compared to a basic treebank grammar (Charniak, 1996), the grammars of highaccuracy parsers weaken independence assumptions by splitting grammar symbols and rules with either lexical (Charniak, 2000; Collins, 1999) or nonlexical (Klein and Manning, 2003; Matsuzaki et al. , 2005) conditioning information. |
| P06-1117 229 176:216 The sentences were processed using Charniaks parser (Charniak, 2000) to generate parse trees automatically. |
| J05-1004 230 7:501 Introduction Robust syntactic parsers, made possible by new statistical techniques (Ratnaparkhi 1997; Collins 1999, 2000; Bangalore and Joshi 1999; Charniak 2000) and by the availability of large, hand-annotated training corpora (Marcus, Santorini, and Marcinkiewicz 1993; Abeille 2003), have had a major impact on the field of natural language processing in recent years. |
| D07-1022 231 9:273 Take, for example, parsing the Wall Street Journal Penn Treebank, a longstanding task for which highly accurate context-free models stabilized by the year 2000 (Collins, 1999; Charniak, 2000). |
| W04-0824 232 39:92 Training and test data, and the Wordnet glosses, were parsed with Charniaks parser (Charniak, 2000). |
| P07-2057 233 10:124 Generally, the syntactic structure of a sentence is represented as a tree, and parsing is carried out by maximizing the likelihood of the tree (Charniak, 2000; Uchimoto et al. , 1999). |
| W00-1303 234 17:181 Decision Trees(Haruno et al. , 1998) and Maximum Entropy models(Ratnaparkhi, 1997; Uchimoto et al. , 1999; Charniak, 2000) have been applied to dependency or syntactic structure analysis. |
| W05-1515 235 219:241 (2004) 89.12 89.10 89.14 89.98 90.22 89.74 Charniak (2000) 90.09 90.01 90.17 89.50 89.69 89.32 bias is to cope with the noise by favoring negative confidences predictions for ambiguous ADJP and ADVP decisions, hence their abysmal labelled recall. |
| W05-1515 236 196:241 3.7 Test Set Results Table 9 shows the results of our best parser on the 15 words test set, as well as the accuracy reported for a recent discriminative parser (Taskar et al. , 2004) and scores we obtained by training and testing the parsers of Charniak (2000) and Bikel (2004) on the same data. |
| W05-1515 237 198:241 Both Charniak (2000) and Bikel (2004) were trained using the goldstandard tags, as this produced higher accuracy on the development set than using Ratnaparkhi (1996)s tags. |
| W06-0901 238 72:160 These sentences are parsed using the August 2005 release of the Charniak parser (Charniak, 2000)4. |
| D07-1027 239 10:285 However, with few exceptions (Model 3 of Collins, 1999; Schmid, 2006), output trees produced by state-of-the-art broad coverage statistical parsers (Charniak, 2000; Bikel, 2004) are only surface context-free phrase structure trees (CFG-trees) without empty categories and coindexation to represent displaced constituents. |
| W03-0402 240 196:239 CH00 = (Charniak, 2000). |
| W03-0402 241 208:239 The performance of our system matches the results of (Charniak, 2000), but is a little lower than the results of the Boosting system in (Collins, 2000), except that the percentage of sentences with no crossing brackets is 1% higher than that of (Collins, 2000). |
| N04-1020 242 58:249 3 Parameter Estimation 3.1 Data Extraction Subordinate clauses (and their main clause counterparts) were extracted from the BLLIP corpus (30 M words), a Treebank-style, machine-parsed version of the Wall Street Journal (WSJ, years 198789) which was produced using Charniaks (2000) parser. |
| P05-1023 243 139:176 For comparison to previous results, table 2 lists the results on the testing set for our best model (TOP-Efficient-Freq20) and several other statistical parsers (Collins, 1999; Collins and Duffy, 2002; Collins and Roark, 2004; Henderson, 2003; Charniak, 2000; Collins, 2000; Shen and Joshi, 2004; Shen et al. , 2003; Henderson, 2004; Bod, 2003). |
| P05-1023 244 65:176 3.1 A History-Based Probability Model As with many other statistical parsers (Ratnaparkhi, 1999; Collins, 1999; Charniak, 2000), Henderson (2003) uses a history-based model of parsing. |
| P03-1054 245 18:233 Charniak (2000) shows the value his parser gains from parentannotation of nodes, suggesting that this information is at least partly complementary to information derivable from lexicalization, and Collins (1999) uses a range of linguistically motivated and carefully hand-engineered subcategorizations to break down wrong context-freedom assumptions of the naive Penn treebank covering PCFG, such as differentiating base NPs from noun phrases with phrasal modifiers, and distinguishing sentences with empty subjects from those where there is an overt subject NP. |
| P03-1054 246 198:233 12This is part of the explanation of why (Charniak, 2000) finds that early generation of head tags as in (Collins, 1999) is so beneficial. |
| P03-1054 247 8:233 In the following decade, great success in terms of parse disambiguation and even language modeling was achieved by various lexicalized PCFG models (Magerman, 1995; Charniak, 1997; Collins, 1999; Charniak, 2000; Charniak, 2001). |
| W04-0308 248 10:145 Parsers that attempt to disambiguate the input completely full parsing typically first employ some kind of dynamic programming algorithm to derive a packed parse forest and then applies a probabilistic top-down model in order to select the most probable analysis (Collins, 1997; Charniak, 2000). |
| P07-1080 249 143:171 It should also be noted that the model (Charniak, 2000) is the most accurate generative model on the standard WSJ parsing benchmark, which confirms the viability of our generative model. |
| P07-1080 250 142:171 This improvement is statically significant.3 The MF model achieves results which do not appear to be significantly different from the results of the best model in the list (Charniak, 2000). |
| P07-1080 251 17:171 (Charniak, 2000; Collins, 1999)). |
| P07-1080 252 140:171 637 ferent generative and discriminative parsing methods (Bikel, 2004; Taskar et al. , 2004; Turian and Melamed, 2006; Charniak, 2000) evaluated in the same experimental setup. |
| P07-1080 253 68:171 Although the most accurate parsing models (Charniak and Johnson, 2005; Henderson, 2004; Collins, 2000) are discriminative, all the most accurate discriminative models make use of a generative model. |
| P07-1080 254 126:171 It is expensive to train R P F1 Bikel, 2004 87.9 88.8 88.3 Taskar et al. , 2004 89.1 89.1 89.1 NN method 89.1 89.2 89.1 Turian and Melamed, 2006 89.3 89.6 89.4 MF method 89.3 90.7 90.0 Charniak, 2000 90.0 90.2 90.1 Table 1: Percentage labeled constituent recall (R), precision (P), combination of both (F1) on the testing set. |
| P07-1080 255 164:171 Both methods are empirically compared, and the mean field approach achieves significantly better results, which are non-significantly different from the results of the most accurate generative parsing model (Charniak, 2000) on our testing set. |
| W03-0312 256 58:158 The Japanese parser outputs the phrasal dependency structure of an input, and that is used as is. We used The Japanese parser KNP (Kurohashi and Nagao, 1994) and The English nl-parser (Charniak, 2000). |
| N06-2026 257 4:89 1 Introduction Recent successes in statistical syntactic parsing based on supervised techniques trained on a large corpus of syntactic trees (Collins, 1999; Charniak, 2000; Henderson, 2003) have brought the hope that the same approach could be applied to the more ambitious goal of recovering the propositional content and the frame semantics of a sentence. |
| N06-2026 258 72:89 These systems all use (Charniak, 2000)s parse trees both for training and testing, as well as various other information sources including sets of n-best parse trees, chunks, or named entities. |
| N06-2026 259 68:89 However, state-of-theart semantic role labelling systems (CoNLL, 2005) use parse trees output by state-of-the-art parsers (Collins, 1999; Charniak, 2000), both for training and testing, and return partial trees annotated with semantic role labels. |
| H05-1001 260 142:277 4.4 Results with GUITAR To use GUITAR, we first parsed the texts using Charniaks parser (Charniak, 2000). |
| N06-1022 261 61:188 (2005) have implemented a dependency parser with good accuracy (it is almost as good at dependency parsing as Charniak (2000)) and very impressive speed (it is about ten times faster than Collins (1997) and four times faster than Charniak (2000)). |
| N06-1022 262 162:188 It has been repeatedly shown to improve parsing accuracy (Johnson, 1998; Charniak, 2000; Klein and Manning, 2003b), but it is difficult if not impossible to integrate with best-first search in bottom-up chart-parsing (as in Charniak et al. |
| N06-1022 263 42:188 The parser of Charniak (2000) is also a two-stage ctf model, where the first stage is a smoothed Markov grammar (it uses up to three previous constituents as context), and the second stage is a lexicalized Markov grammar with extra annotations about parents and grandparents. |
| N03-1024 264 77:235 Given a sentence group, we pass each of the 11 sentences to Charniaks (2000) parser to get 11 parse trees. |
| W01-0712 265 204:210 The preliminary results are encouraging though not as good as advanced statistical parsers like those of Charniak (2000) and Collins (2000). |
| C04-1180 266 5:129 1 Introduction The levels of accuracy and robustness recently achieved by statistical parsers (e.g. Collins (1999), Charniak (2000)) have led to their use in a number of NLP applications, such as question-answering (Pasca and Harabagiu, 2001), machine translation (Charniak et al. , 2003), sentence simplification (Carroll et al. , 1999), and a linguists search engine (Resnik and Elkiss, 2003). |
| P03-2012 267 66:175 The texts were parsed by E. Charniaks parser (Charniak, 2000). |
| P06-1055 268 11:234 Therefore, a variety of techniques have been developed to both enrich and generalize the naive grammar, ranging from simple tree annotation and symbol splitting (Johnson, 1998; Klein and Manning, 2003) to full lexicalization and intricate smoothing (Collins, 1999; Charniak, 2000). |
| P06-1055 269 8:234 1 Introduction Probabilistic context-free grammars (PCFGs) underlie most high-performance parsers in one way or another (Collins, 1999; Charniak, 2000; Charniak and Johnson, 2005). |
| P07-1035 270 18:171 Hence, state-of-the-art parsers either supplement the part-of-speech (POS) tags with the lexical forms themselves (Collins, 2003; Charniak, 2000), manually split the tagset into a finer-grained one (Klein and Manning, 2003a), or learn finer grained tag distinctions using a heuristic learning procedure (Petrov et al. , 2006). |
| C08-1050 271 55:214 A constituent-based system using Charniaks parser (Charniak, 2000). |
| C08-1050 272 16:214 By habit, most systems for automatic role-semantic analysis have used Pennstyle constituents (Marcus et al., 1993) produced by Collins (1997) or Charniaks (2000) parsers. |
| P02-1055 273 13:174 Charniak (2000) notes that having his generative parser generate the POS of a constituents head before the head itself increases performance by 2 points. |
| P04-1087 274 91:214 These sentences were then parsed using a statistical parser (Charniak, 2000). |
| W05-0638 275 24:90 In CoNLL-2005, full parsing trees are provided by two full parsers: the Collins parser (Collins, 1999) and the Charniak parser (Charniak, 2000). |
| C04-1021 276 75:219 2.1 Parser Enhancements We used the Charniak parser (Charniak, 2000) for the experiments reported in this paper. |
| E06-1038 277 114:236 (2005b) parser and phrase structure tree from the Charniak (2000) parser. |
| E06-1038 278 8:236 We focus on the particular instantiation of sentence compression when the goal is to produce the compressed version solely by removing words or phrases from the original, which is the most common setting in the literature (Knight and Marcu, 2000; Riezler et al. , 2003; Turner and Charniak, 2005). |
| E06-1038 279 110:236 To do this we parse every sentence twice, once with a dependency parser (McDonald et al. , 2005b) and once with a phrase-structure parser (Charniak, 2000). |
| P06-2089 280 9:169 Some of the more popular and more accurate of these approaches to data-driven parsing (Charniak, 2000; Collins, 1997; Klein and Manning, 2002) have been based on generative models that are closely related to probabilistic contextfree grammars. |
| P06-2089 281 148:169 We tested this hypothesis by using the Charniak (2000) parser in n-best mode, producing the top 10 trees with corresponding probabilities. |
| P06-2089 282 141:169 In the probabilistic LR model, probabilities are assigned to tree 696 Precision Recall F-score Time (min) Best-First Classifier-Based (this paper) 88.1 87.8 87.9 17 Deterministic (MaxEnt) (this paper) 85.4 84.8 85.1 < 1 Charniak & Johnson (2005) 91.3 90.6 91.0 Unk Bod (2003) 90.8 90.7 90.7 145* Charniak (2000) 89.5 89.6 89.5 23 Collins (1999) 88.3 88.1 88.2 39 Ratnaparkhi (1997) 87.5 86.3 86.9 Unk Tsuruoka & Tsujii (2005): deterministic 86.5 81.2 83.8 < 1* Tsuruoka & Tsujii (2005): search 86.8 85.0 85.9 2* Sagae & Lavie (2005) 86.0 86.1 86.0 11* Table 1: Summary of results on labeled precision and recall of constituents, and time required to parse the test set. |
| P06-2089 283 140:169 A probabilistic shiftreduce LR-like model, such as the one used in our parser, is different in many ways from a lexicalized PCFG-like model (using markov a grammar), such as those used in the Collins (1999) and Charniak (2000) parsers. |
| P06-2089 284 153:169 See (Charniak, 2000) for details. |
| W08-0406 285 106:250 The English side is parsed using a state-of-the-art statistical English parser (Charniak, 2000). |
| W08-1005 286 9:154 Therefore, a variety of techniques have been developed to both enrich and generalize the naive grammar, ranging from simple tree annotation and symbol splitting (Johnson, 1998; Klein and Manning, 2003) to full lexicalization and intricate smoothing (Collins, 1999; Charniak, 2000). |
| W08-1005 287 6:154 1 Introduction Probabilistic context-free grammars (PCFGs) underlie most high-performance parsers in one way or another (Collins, 1999; Charniak, 2000; Charniak and Johnson, 2005). |
| P05-1038 288 159:222 (1999) for Czech, where the bigram model upped dependency accuracy by about 0.9%, as well as for English where Charniak (2000) reports an increase in F-score of approximately 0.3%. |
| E06-1011 289 18:205 Previous work has shown that conditioning on neighboring decisions can lead to significant improvements in accuracy (Yamada and Matsumoto, 2003; Charniak, 2000). |
| E06-1011 290 48:205 The score of a tree for secondorder parsing is now s(x,y) = summationdisplay (i,k,j)y s(i,k,j) where k and j are adjacent, same-side children of i in the tree y. The second-order model allows us to condition onthe mostrecent parsing decision, thatis, the last dependent picked up by a particular word, which is analogous to the the Markov conditioning of in the Charniak parser (Charniak, 2000). |
| P06-1022 291 87:186 Incidentally, the above attributes are the same as those used by the conventional stochastic dependency parsing methods (Collins, 1996; Ratnaparkhi, 1997; Fujio and Matsumoto, 1998; Uchimoto et al. , 1999; Charniak, 2000; Kudo and Matsumoto, 2002). |
| W07-1527 292 84:168 The resulting English corpus contained 10,000 sentences which were syntactically parsed (Charniak, 2000). |
| C04-1040 293 22:230 However, the accuracy of Yamadas analyzer is lower than state-of-the-art phrase structure parsers such as Charniaks Maximum-Entropy-Inspired Parser (MEIP) (Charniak, 2000) and Collins Model 3 parser. |
| W06-1603 294 75:249 A pair of sentences is rst fed to a syntactic parser (Charniak, 2000) and then passed to a semantic role labeler (ASSERT; (Pradhan et al. , 2004)), to label predicate argument tuples. |
| P03-1069 295 97:237 The corpus is distributed in a Treebankstyle machine-parsed version which was produced with Charniaks (2000) parser. |
| P05-1012 296 5:209 The best phrase-structure parsing models represent generatively the joint probability P(x,y) of sentence x having the structure y (Collins, 1999; Charniak, 2000). |
| P05-1012 297 174:209 3.1 Lexicalized Phrase Structure Parsers It is well known that dependency trees extracted from lexicalized phrase structure parsers (Collins, 1999; Charniak, 2000) typically are more accurate than those produced by pure dependency parsers (Yamada and Matsumoto, 2003). |
| P06-2038 298 102:105 299 4 Future Work While the modi cation given in section 2.2 is speci c to CYK parsing, we believe that placing restrictions based on the output of a chunk parser is general enough to be applied to any generative, statistical parser, such as the Charniak parser (2000), or a Lexical Tree Adjoining Grammar based parser (Sarkar, 2000). |
| P06-1033 299 55:272 Thus, Nivre and Nilsson (2005) improve parsing accuracy for MaltParser by projectivizing training data and applying an inverse transformation to the output of the parser, while Hall and Novak (2005) apply post-processing to the output of Charniaks parser (Charniak, 2000). |
| P06-1048 300 157:244 RASP failed on 17 sentences from the Broadcast news corpus and 33 from the Ziff-Davis corpus; Charniaks (2000) parser successfully parsed the Broadcast News corpus but failed on three sentences from the Ziff-Davis corpus. |
| P06-1048 301 17:244 Many algorithms exploit parallel corpora (Jing 2000; Knight and Marcu 2002; Riezler et al. 2003; Nguyen et al. 2004a; Turner and Charniak 2005; McDonald 2006) to learn the correspondences between long and short sentences in a supervised manner, typically using a rich feature space induced from parse trees. |
| P06-1048 302 145:244 The training data for both models was parsed using Charniaks (2000) parser. |
| P06-1048 303 102:244 Ziff-Davis Corpus Most previous work (Jing 2000; Knight and Marcu 2002; Riezler et al. 2003; Nguyen et al. 2004a; Turner and Charniak 2005; McDonald 2006) has relied on automatically constructed parallel corpora for training and evaluation purposes. |
| W06-2611 304 173:216 The sentences were processed using Charniaks parser (Charniak, 2000) to generate parse trees automatically. |
| P07-2052 305 7:109 1 Introduction In recent years, many accurate phrase-structure parsers have been developed (e.g. , (Collins, 1999; Charniak, 2000)). |
| I08-1016 306 57:147 In our implementation, a context free grammar probabilistic parser (Charniak, 2000) was used to parse the input. |
| W03-1022 307 120:201 We parsed the WordNet definitions and example sentences with the same syntactic parser used for Bllip (Charniak, 2000). |
| P07-1025 308 85:217 For this task we utilized the August 2005 release of the Charniak parser with the default speed/accuracy settings (Charniak, 2000), which required roughly 360 hours of processor time on a 2.5 GHz PowerPC G5. |
| P07-1032 309 38:189 Preiss (2003) compares the parsers of Collins (2003) and Charniak (2000), the GR finder of Buchholz et al. |
| P07-1032 310 8:189 1 Introduction Parsers have been developed for a variety of grammar formalisms, for example HPSG (Toutanova et al. , 2002; Malouf and van Noord, 2004), LFG (Kaplan et al. , 2004; Cahill et al. , 2004), TAG (Sarkar and Joshi, 2003), CCG (Hockenmaier and Steedman, 2002; Clark and Curran, 2004b), and variants of phrase-structure grammar (Briscoe et al. , 2006), including the phrase-structure grammar implicit in the Penn Treebank (Collins, 2003; Charniak, 2000). |
| D08-1091 311 11:223 In recent years, latent variable methods have been shown to produce grammars which are as good as, or even better than, earlier parsing work (Collins, 1999; Charniak, 2000). |
| W03-1707 312 12:178 The creation of the Penn English Treebank (Marcus et al. , 1993), a syntactically interpreted corpus, played a crucial role in the advances in natural language parsing technology (Collins, 1997; Collins, 2000; Charniak, 2000) for English. |
| W02-1009 313 30:288 Most researchers have used an n-gram model (Eisner, 1996; Charniak, 2000) or more general Markov model (Alshawi, 1996) to model the sequence of nonterminals in the RHS. |
| W06-1622 314 60:165 It includes several Wall Street Journal sections with parse-trees from both Charniaks (2000) parser and Collins (1999) parser. |
| W04-0305 315 71:189 3 A Generative Left-Corner Probability Model As with several previous statistical parsers (Collins, 1999; Charniak, 2000), we use a generative history-based probability model of parsing. |
| C08-1069 316 114:172 We compared the Enju parser with four CFG parsers: Stanfords lexicalized parser (Klein and Manning, 2003), Collins parser (Collins, 1999), Charniaks parser (Charniak, 2000), and Charniak and Johnsons reranking parser (Charniak and Johnson, 2005). |
| P03-1046 317 4:174 1 Introduction State-of-the-art statistical parsers for Penn Treebank-style phrase-structure grammars (Collins, 1999), (Charniak, 2000), but also for Categorial Grammar (Hockenmaier and Steedman, 2002b), include models of bilexical dependencies defined in terms of local trees. |
| P03-1046 318 51:174 Therefore, the head words of arguments (such as Smith) are generated in the following manner: C8B4DB CP CYCR CP BNCWCWCR CW BNDB CW CXBNCXBNCWCR CP BNDB CP CXCXB5 The head word of modifiers (such as yesterday)are generated differently: C8B4DB D1 CYCR D1 BNCWCWCR D1 BNDB D1 CXBNCXBNCWCR CW BNDB CW CXB5 Like Collins (1999) and Charniak (2000), the SD model assumes that word-word dependencies can be defined at the maximal projection of a constituent. |
| C04-1204 319 33:135 However, the accuracy of those parsers was still below PCFG parsers (Collins, 1999; Charniak, 2000) in terms of the PARSEVAL score, i.e., labeled bracketing accuracy of CFG-style parse trees. |
| C04-1204 320 6:135 However, their accuracy was still below the state-of-theart PCFG parsers (Collins, 1999; Charniak, 2000) in terms of the PARSEVAL score. |
| N04-2009 321 27:135 The documents are then parsed using Eugene Charniaks maximum entropy inspired parser (Charniak, 2000). |
| N04-1011 322 19:140 Even though this seems linguistically highly unnatural (e.g. , punctuation might indicate suprasegmental prosodic properties), statistical parsers generally perform signi cantly better when their training and test data contains punctuation represented in this way than if the punctuation is stripped out of the training and test data (Charniak, 2000; Engel et al. , 2002; Johnson, 1998). |
| P07-1122 323 36:184 (1999) and Charniak (2000) adapted to Czech are not able to create the non-projective arcs present in the treebank, which is unsatisfactory. |
| P07-1122 324 7:184 This can be seen in state-of-the-art constituency-based parsers such as Collins (1999), Charniak (2000), and Petrovetal.(2006),andtheeffectsofdifferenttransformations have been studied by Johnson (1998), KleinandManning(2003),andBikel(2004). |
| P05-1011 325 114:172 Different from the results of CCG and PCFG (Collins, 1999; Charniak, 2000), the recall was clearly lower than precision. |
| P04-1042 326 97:176 In Table 2 we present comparative results, using the PARSEVAL-based evaluation metric introduced by Johnson (2002) a correct empty category inference requires the string position of the empty category, combined with the left and right boundaries plus syntactic category of the antecedent, if any, for purposes of comparison.9,10 Note that this evaluation metric does not require correct attachment of the empty category into 8A complete description of feature templates can be found at http://nlp.stanford.edu/rog/acl2004/templates/index.html 9For purposes of comparability with Johnson (2002) we used Charniaks 2000 parser as P. 10Our algorithm was evaluated on a more stringent standard for NP-* than in previous work: control loci-related mappings were done after dislocated nodes were actually relocated by the algorithm, so an incorrect dislocation remapping can render incorrect the indices of a correct NP-* labeled bracketing. |
| P04-1042 327 137:176 Note that the dependency figures of Dienes lag behind even the parsed results for Johnsons model; this may well be due to the fact that Dienes built his model as an extension of Collins (1999), which lags behind Charniak (2000) by about 1.3-1.5%. |
| P04-1042 328 126:176 State-of-the-art statistical parsing is far better on WSJ (Charniak, 2000) than on NEGRA (Dubey and Keller, 2003), so for comparison of parser-composed dependency performance we used vanilla PCFG models for both WSJ and NEGRA trained on comparably-sized datasets; in addition to making similar types of independence assumptions, these models performed relatively comparably on labeled bracketing measures for our development sets (73.2% performance for WSJ versus 70.9% for NEGRA). |
| W08-2226 329 30:118 For instance, the introductionof large-scalelexicaland syntacticresources likethePennTreeBank(Marcusetal.,1993)haveledto highlyaccurate,domainindependent parsers (Collins, 1999; Charniak, 2000). |
| W08-2226 330 22:118 TEXTCAP first uses the domain-independent Charniak parser (Charniak, 2000) to convert sentences in the source document into a sequence of syntactic parses. |
| C08-1144 331 57:207 Each target sentence in the training corpus is parsed with a stochastic parserwe use Charniak (2000))to produce constituent labels for target spans. |
| C02-1159 332 23:252 Using existing parsers such as (Charniak, 2000; Collins, 1999) would eliminate the need to build a parser from scratch. |
| W05-0639 333 22:86 In order to attack this problem, in addition to Charniaks parser (Charniak, 2000), our system combine two parser which are trained on both syntactic constituent information and semantic argument information. |
| W05-0639 334 19:86 2.1 Parsing Previous SRL systems usually use a pure syntactic parser, such as (Charniak, 2000; Collins, 1999), to retrieve possible constituents. |
| W06-1668 335 14:239 To avoid this problem, generative models for NLP tasks have often been manually designed to achieve an appropriate representation of the joint distribution, such as in the parsing models of (Collins, 1997; Charniak, 2000). |
| W06-1668 336 11:239 Additionally, many discriminative models use a generative model as a base model and add discriminative features with reranking (Collins, 2000; Charniak and Johnson, 2005; Roark et al. , 2004), or train discriminatively a small set of weights for features which are generatively estimated probabilities (Raina et al. , 2004; Och and Ney, 2002). |
| D07-1072 337 25:303 Lexical methods split each pre-terminal symbol into many subsymbols, one for each word, and then focus on smoothing sparse lexical statis688 tics (Collins, 1999; Charniak, 2000). |
| P05-1022 338 2:180 c2005 Association for Computational Linguistics Coarse-to-fine n-best parsing and MaxEnt discriminative reranking Eugene Charniak and Mark Johnson Brown Laboratory for Linguistic Information Processing (BLLIP) Brown University Providence, RI 02912 {mj|ec}@cs.brown.edu Abstract Discriminative reranking is one method for constructing high-performance statistical parsers (Collins, 2000). |
| P05-1022 339 56:180 Things become worse still in a parser like the one described in Charniak (2000) because it conditions on (and hence splits the dynamic programming states according to) features of the grandparent node in addition to the parent, thus multiplying the number of possible dynamic programming states even more. |
| P05-1022 340 104:180 Finally, we note that 50-best parsing is only a fac1Charniak in (Charniak, 2000) cites an accuracy of 89.5%. |
| P05-1022 341 8:180 The 50-best parser is a probabilistic parser that on its own produces high quality parses; the maximum probability parse trees (according to the parsers model) have an f-score of 0.897 on section 23 of the Penn Treebank (Charniak, 2000), which is still state-of-the-art. |
| P05-1022 342 4:180 This paper describes a simple yet novel method for constructing sets of 50-best parses based on a coarse-to-fine generative parser (Charniak, 2000). |
| N06-1040 343 51:188 One common strategy in statistical parsing is what can be termed an approximate coarse-to-fine approach: a simple PCFG is used to prune the search space to which richer and more complex models are applied subsequently (Charniak, 2000; Charniak and Johnson, 2005). |
| N06-1040 344 179:188 niak parser (Charniak, 2000; Charniak and Johnson, 2005) uses a Markov order-3 baseline PCFG in the initial pass, with a best-first algorithm that is run past the first parse to populate the chart for use by the richer model. |
| N06-1040 345 31:188 Probability estimates of the RHS given the LHS are often smoothed by making a Markov assumption regarding the conditional independence of a category on those more than k categories away (Collins, 1997; Charniak, 2000): P(X Y1Yn)= P(Y1|X) nY i=2 P(Yi|X,Y1 Yi1) P(Y1|X) nY i=2 P(Yi|X,Yik Yi1). |
| N03-1027 346 115:172 Model interpolation in this case perSystem Training Heldout LR LP MAP Brown;T Brown;H 76.0 75.4 MAP Brown;T WSJ;24 76.9 77.1 Gildea WSJ;2-21 86.1 86.6 MAP WSJ;2-21 WSJ;24 86.9 87.1 Charniak (1997) WSJ;2-21 WSJ;24 86.7 86.6 Ratnaparkhi (1999) WSJ;2-21 86.3 87.5 Collins (1999) WSJ;2-21 88.1 88.3 Charniak (2000) WSJ;2-21 WSJ;24 89.6 89.5 Collins (2000) WSJ;2-21 89.6 89.9 Table 4: Parser performance on WSJ;23, baselines. |
| N03-1027 347 63:172 The PCFG is a Markov grammar (Collins, 1997; Charniak, 2000), i.e. the production probabilities are estimated by decomposing the joint probability of the categories on the right-hand side into a product of conditionals via the chain rule, and making a Markov assumption. |
| C00-1011 348 142:160 These scores are higher than those of several other parsers (e.g. Collins 1997, 99; Charniak 1997), but remain behind tim scores of Charniak (2000) who obtains 90.1% LP and 90.1% LR for sentences _< 40 words. |
| C00-1011 349 126:160 40,000 sentences) and section 23 for testing (see Collins 1997, 1999; Charniak 1997, 2000; l~,atnalmrkhi 1999); we only tested on sentences _< 40 words (2245 sentences). |
| W05-1511 350 77:178 They include grammar compilation (Tomita, 1986; Nederhof, 2000), the Viterbi algorithm, controlling search strategies without FOM such as left-corner parsing (Rosenkrantz and Lewis II, 1970) or headcorner parsing (Kay, 1989; van Noord, 1997), and with FOM such as the beam search, the best-first search or A* search (Chitrao and Grishman, 1990; Caraballo and Charniak, 1998; Collins, 1999; Ratnaparkhi, 1999; Charniak, 2000; Roark, 2001; Klein and Manning, 2003). |
| P04-1006 351 136:172 on the BLLIP99 corpus (Charniak et al. , 1999); a corpus of 30million words automatically parsed using the Charniak parser (Charniak, 2000). |
| P04-1006 352 46:172 2.1 nbest list reranking Much effort has been put forth in developing efficient probabilistic models for parsing strings (Caraballo and Charniak, 1998; Goldwater et al. , 1998; Blaheta and Charniak, 1999; Charniak, 2000; Charniak, 2001); an obvious solution to parsing wordlattices is to use nbest list reranking. |
| P04-1006 353 59:172 In Figure 4 we present the general overview of a multi-stage parsing technique (Goodman, 1997; Charniak, 2000; Charniak, 2001). |
| W05-0623 354 67:79 Recent releases of the Charniak parser (Charniak, 2000) have included an option to provide the top k parses of a given sentence according to the probability model of the parser. |
| W06-2937 355 20:149 2 System Description 241 Over the past decades, many state-of-the-art parsing algorithm were proposed, such as head-word lexicalized PCFG (Collins, 1998), Maximum Entropy (Charniak, 2000), Maximum/Minimum spanning tree (MST) (McDonald et al. , 2005), Bottom-up deterministic parsing (Yamada and Matsumoto, 2003), and Constant-time deterministic parsing (Nivre, 2003). |
| C04-1097 356 52:154 (24 for German, 23 for French) We denote that set of features in shorthand as () i f . With this extension, a model of Markov 3 A Markov grammar is a model of constituent structure that starts at the root of the tree and assigns probability to the expansion of a non-terminal one daughter at a time, rather than as entire productions (Charniak, 1997 & 2000). |
| C04-1097 357 150:154 Experiments by Daum et al (2002) and the parsing work of Charniak (2000) and others indicate that further lexicalization may yield some additional improvements for ordering. |
| C04-1097 358 124:154 The PTB to DSIF transformation pipeline includes the following stages, inspired by Langkilde-Gearys (2002b) description: A. Deserialize the tree B. Label heads, according to Charniaks head labeling rules (Charniak, 2000) C. Remove empty nodes and flatten any remaining empty non-terminals D. Relabel heads to conform more closely to the head conventions of NLPWin E. Label with logical roles, inferred from PTB functional roles F. Flatten to maximal projections of heads (MPH), except in the case of conjunctions G. Flatten non-branching non-terminals H. Perform dictionary look-up and morphological analysis I. Introduce structure for material between paired delimiters and for any coordination not already represented in the PTB J. Remove punctuation K. Remove function words L. Map the head of each maximal projection to a dependency node, and map the heads modifiers to the first nodes dependents, thereby forming a complete dependency tree. |
| J04-3001 359 10:405 Current state-of-the-art statistical parsers (Collins 1999; Charniak 2000) are all trained on large annotated corpora such as the Penn Treebank (Marcus, Santorini, and Marcinkiewicz 1993). |
| W07-2085 360 62:108 (Charniak, 2000)). |
| N07-1072 361 59:189 3.3 Preprocessing for Parsing We first used the Charniak parser (2000) to parse the original skill statements. |
| W00-1201 362 7:100 The success of statistical methods in particular has been quite evident in the area of syntactic parsing, most recently with the outstanding results of (Charniak, 2000) and (Colhns, 2000) on the now-standard English test set of the Penn Treebank (Marcus et al. , 1993). |
| P08-1013 363 7:210 More accurate statistical models of natural language have mainly been developed in the field of statistical parsing, e.g. Collins (2003), Charniak (2000) and Ratnaparkhi (1999). |
| N06-2019 364 5:84 Reported experiments measure the effect of early deletion under in-domain and out-of-domain parser training conditions using a state-of-the-art parser (Charniak, 2000). |
| N06-2019 365 33:84 3 Experiments This section reports parsing experiments studying the effect of early deletion under in-domain and outof-domain parser training conditions using the August 2005 release of the Charniak parser (2000). |
| N06-2019 366 50:84 The dependency metric performs syntactic head-matching for each word using a set of given head percolation rules (derived from Charniaks parser (2000)), and its relaxed formulation ignores terminals spanned by FILLER and EDITED constituents. |
| N06-2019 367 24:84 Experiments performed (3.3) use a state-of-the-art parser (Charniak, 2000) to study the impact of early filler deletion under in-domain and out-of-domain (i.e. adaptation) training conditions. |
| W06-2002 368 4:162 1 Introduction Much of the current research into probabilistic parsing is founded on probabilistic contextfreegrammars(PCFGs)(Collins,1999;Charniak, 2000; Charniak, 2001). |
| P08-1037 369 30:210 Parsing As our baseline parsers, we use two state-of-theart lexicalised parsing models, namely the Bikel parser (Bikel, 2004) and Charniak parser (Charniak, 2000). |
| P08-1037 370 9:210 For example, a number of different parsers have been shown to benet from lexicalisation, that is, the conditioning of structural features on the lexical head of the given constituent (Magerman, 1995; Collins, 1996; Charniak, 1997; Charniak, 2000; Collins, 2003). |
| W06-2305 371 12:167 Compared to traditional handwritten grammars, successfulstochasticmodelslike(Collins,1999;Charniak, 2000) open up an even greater space of alternatives for the parser and accordingly offer a great deal of opportunities to construct odd structural descriptions from them. |
| P05-1039 372 145:187 This is worrying: at times in the literature, details of search or smoothing are left out (e.g. Charniak (2000)). |
| P05-1039 373 46:187 The first is a Markov context-free rule (Magerman, 1995; Charniak, 2000). |
| P05-1039 374 85:187 If the tags are necessary for semantic interpretation, presumably they could be re-inserted using a strategy such as that of Blaheta and Charniak (2000) The last transformation therefore removes the GF of S nodes. |
| W00-1308 375 174:186 This presumably corresponds with Charniak's (2000: 136) observation that Section 23 of the Penn Treebank is easier than some others. |
| W05-0620 376 138:334 Test WSJ Test Brown P(%) R(%) F1 P(%) R(%) F1 P(%) R(%) F1 UPC Chunker 94.66 93.17 93.91 95.26 94.52 94.89 92.64 90.85 91.73 UPC Clauser 90.38 84.73 87.46 90.93 85.94 88.36 84.21 74.32 78.95 Collins (1999) 85.02 83.55 84.28 85.63 85.20 85.41 82.68 81.33 82.00 Charniak (2000) 87.60 87.38 87.49 88.20 88.30 88.25 80.54 81.15 80.84 Table 3: Results of the syntactic parsers on the development, and WSJ and Brown test sets. |
| W05-0620 377 111:334 Full parser of Charniak (2000). |
| W05-0620 378 123:334 tWSJ tBrown UPC PoS-tagger 97.13 97.36 94.73 Charniak (2000) 92.01 92.29 87.89 Table 2: Accuracy (%) of PoS taggers. |
| P08-1067 379 167:181 type system F1% D Collins (2000) 89.7 Henderson (2004) 90.1 Charniak and Johnson (2005) 91.0 updated (Johnson, 2006) 91.4 this work 91.7 G Bod (2003) 90.7Petrov and Klein (2007) 90.1 S McClosky et al. |
| P08-1067 380 47:181 Following (Charniak and Johnson, 2005), the first feature f1(y) = logPr(y) is the log probability of a parse from the baseline generative parser, while the remaining features are all integer valued, and each of them counts the number of times that a particular configuration occurs in parse y. For example, one such feature f2000 might be a question how many times is a VP of length 5 surrounded by the word has and the period? |
| P08-1067 381 39:181 However, in this work, we use forests from a Treebank parser (Charniak, 2000) whose grammar is often flat in many productions. |
| P08-1067 382 49:181 Using a machine learning algorithm, the weight vector w can be estimated from the training data where each sentence si is labelled with its correct (gold-standard) parse yi . As for the learner, Collins (2000) uses the boosting algorithm and Charniak and Johnson (2005) use the maximum entropy estimator. |
| P08-1067 383 125:181 Those with a are from (Collins, 2000), and others are from (Charniak and Johnson, 2005), with simplifications. |
| P08-1067 384 130:181 Our feature set is summarized in Table 2, which closely follows Charniak and Johnson (2005), except that we excluded the non-local features Edges, NGram, and CoPar, and simplified Rule and NGramTree features, since they were too complicated to compute.4 We also added four unlexicalized local features from Collins (2000) to cope with data-sparsity. |
| C02-1126 385 23:139 criminative models described in (Magerman, 1995; Ratnaparkhi, 1997), the lexicalized PCFG models in (Collins, 1999), the generative model in (Charniak, 2000), the lexicalized TAG extractor in (Xia, 1999) and the stochastic lexicalized TAG models in (Chiang, 2000; Sarkar, 2001; Chen and VijayShanker, 2000). |
| W08-1112 386 41:133 Methodologies such as lexicalisation (Collins, 1997; Charniak, 2000) and tree transformations (Johnson, 1998), weaken the independence assumptions and have been applied successfully to parsing and shown significant improvements over simple PCFGs. |
| W08-1112 387 43:133 3.1 A History-Based Model The history-based (HB) approach which incorporates more context information has worked well in parsing (Collins, 1997; Charniak, 2000). |
| D07-1028 388 97:188 In this way, the generation model resembles history-based models for parsing (Black et al. , 1992; Collins, 1999; Charniak, 2000). |
| D07-1028 389 13:188 This history-based approach has worked well in parsing (Collins, 1999; Charniak, 2000) and we show that it also improves PCFG-based generation. |
| P06-2010 390 127:184 Train Devel tWSJ tBrown Sentences 39,832 1,346 2,416 426 Tokens 950,028 32,853 56,684 7,159 Propositions 90,750 3,248 5,267 804 Arguments 239,858 8,346 14,077 2,177 Table 2: Counts on the data set The preprocessing modules used in CONLL2005 include an SVM based POS tagger (Gimenez and M`arquez, 2003), Charniak (2000)s full syntactic parser, and Chieu and Ng (2003)s Named Entity recognizer. |
| P07-1103 391 123:235 Each of these variants was then submitted to a parser trained on written text (Charniak, 2000). |
| P05-1025 392 4:182 By using a statistical parser (Charniak, 2000) and memorybased learning tools for classification (Daelemans et al. , 2004), we obtain high precision and recall of several GRs. |
| P05-1025 393 50:182 This dependency extraction procedure from constituent trees gives us a straightforward way to obtain unlabeled dependencies: use an existing statistical parser (Charniak, 2000) trained on the Penn Treebank to produce constituent trees, and extract unlabeled dependencies using the aforementioned head-finding rules. |
| P07-1078 394 8:197 1 Introduction State of the art statistical parsers (Collins, 1999; Charniak, 2000; Koo and Collins, 2005; Charniak and Johnson, 2005) are trained on manually annotated treebanks that are highly expensive to create. |
| W03-2008 395 9:174 Broad coverage syntactic parsers with good performance have recently become available (Charniak, 2000; Collins, 2000), but they are not trained for patents. |
| C04-1157 396 92:133 5.1 Parsers used Charniaks parser (2000) is a combination probabilistic context free grammar and maximum entropy parser. |
| P06-3004 397 10:199 For the English WSJ, high accuracy parsing models have been created, some of them using extensions to classical PCFG parsing such as lexicalization and markovization (Collins, 1999; Charniak, 2000; Klein and Manning, 2003). |
| P02-1034 398 141:185 (Charniak 2000) describes a different method which achieves very similar performance to (Collins 2000). |
| C04-1010 399 11:293 On the other hand, the best available parsers trained on the Penn Treebank, those of Collins (1997) and Charniak (2000), use statistical models for disambiguation that make crucial use of dependency relations. |
| C04-1010 400 93:293 This permits us to make exact comparisons with the parser of Yamada and Matsumoto (2003), but also the parsers of Collins (1997) and Charniak (2000), which are evaluated on the same data set in Yamada and Matsumoto (2003). |
| C04-1010 401 90:293 used for training and section 23 for testing (Collins, 1999; Charniak, 2000). |
| C04-1010 402 120:293 Table 2 shows the dependency accuracy, root accuracy and complete match scores for our best parser (Model 2 with label set B) in comparison with Collins (1997) (Model 3), Charniak (2000), and Yamada and Matsumoto (2003).5 It is clear that, with respect to unlabeled accuracy, our parser does not quite reach state-of-the-art performance, even if we limit the competition to deterministic methods such as that of Yamada and Matsumoto (2003). |
| C04-1010 403 12:293 Moreover, the deterministic dependency parser of Yamada and Matsumoto (2003), when trained on the Penn Treebank, gives a dependency accuracy that is almost as good as that of Collins (1997) and Charniak (2000). |
| C04-1010 404 134:293 In another study, Blaheta and Charniak (2000) report an F-measure of 98.9% for the assignment of Penn Treebank grammatical role labels (our G set) to phrases that were correctly parsed by the parser described in Charniak (2000). |
| W08-0308 405 125:159 The translation is from English to Chinese, and Charniak (2000)s parser, trained on the Penn Treebank, is used to generate the syntax trees for the English side. |
| P04-1030 406 215:232 We achieve 73.2/76.5% LP/LR on section 23 of the Penn Treebank, compared to 82.9/82.4% LP/LR of Roark (2001) and 90.1/90.1% LP/LR of Charniak (2000). |
| P04-1030 407 197:232 The current best-performing models, in terms of WER, for the HUB-1 corpus, are the models of Roark (2001), Charniak (2001) (applied to n-best lists by Hall and Johnson (2003)), and the SLM of Chelba and Jelinek (2000) (applied to n-best lists by Xu et al. |
| P04-1030 408 204:232 Model n-best List/Lattice Training Size WER (%) SER (%) Oracle (50-best lattice) Lattice 7.8 Charniak (2001) List 40M 11.9 Xu (2002) List 20M 12.3 Roark (2001) (with EM) List 2M 12.7 Hall (2003) Lattice 30M 13.0 Chelba (2000) Lattice 20M 13.0 Current ( a1 1a6 16a0 a1 1) List 20M 13.1 71.0 Current ( a1 1a6 16a0 a1 1) Lattice 20M 13.1 70.4 Roark (2001) (no EM) List 1M 13.4 Lattice Trigram Lattice 40M 13.7 69.0 Current ( a1 1a6 16a0 a1 1) List 1M 14.8 74.3 Current ( a1 1a6 16a0 a1 1) Lattice 1M 14.9 74.0 Current ( a1 a1 0) Lattice 1M 16.0 75.5 Treebank Trigram Lattice 1M 16.5 79.8 No language model Lattice 16.8 84.0 Table 3: Comparison of WER for parsing HUB-1 words lattices with best results of other works. |
| W07-1220 409 79:193 The manually compiled grammars in our experiment are also intrinsically different to grammars automatically induced from treebanks (e.g. that used in the Charniak parser (Charniak, 2000) or the various CCG parsers (Hockenmaier, 2006)). |
| E06-1012 410 7:166 With the emergence of the important role of word-to-word relations in parsing (Charniak, 2000; Collins, 1996), dependency grammars have gained acertain popularity; e.g., Yamada and Matsumoto (2003) for English, Kudo and Matsumoto (2000; 2002), Sekine et al. |
| H05-1035 411 20:147 The feature set contains complex information extracted automatically from candidate syntax trees generated by parsing (Charniak, 2000), trees that will be improved by more accurate PP-attachment decisions. |
| P05-1065 412 105:179 The parse features are generated using the Charniak parser (Charniak, 2000) trained on the standard Wall Street Journal Treebank corpus. |
| P05-1065 413 67:179 We also use a standard statistical parser (Charniak, 2000) to provide syntactic analysis. |
| D08-1050 414 79:252 477 Lease and Charniak (2005) obtained an improvement in the accuracy of the Charniak (2000) parser, as well as POS tagging accuracy, when applied to the biomedical domain, by training a new POS tagger model with a combination of newspaper and biomedical data. |
| D07-1065 415 137:188 We evaluate gap insertion on gold trees from section 23 of the Wall Street Journal Penn Treebank (WSJ) and parse trees automatically produced using the Charniak (2000) and Bikel (2004) parsers. |
| N03-1030 416 133:187 Recall Precision F-score a174a113a66a4a112a65a147 28.2 37.1 32.0 a174a74a145a32a112a65a147 25.4 64.9 36.5 a112a113a18a20a11a29a112a170a147 77.1 83.3 80.1 a147a137a3a22a5a117a112a65a147a49a21a95a15a177a183a173a24 82.7 83.5 83.1 a147a62a3a6a5a117a112a170a147a49a21a95a15a44a184a144a24 85.4 84.1 84.7 a149a88a112a170a147 98.2 98.5 98.3 Table 1: Discourse segmenter evaluation Table 1 shows the results obtained by the algorithm described in this paper (a147a62a3a6a5a117a112a170a147a49a21a95a15a104a183a185a24 ) using syntactic trees produced by Charniaks parser (2000), in comparison with the results obtained by the algorithm described in (Marcu, 2000) (a112a113a18a20a11a29a112a170a147 ), and baseline algorithms a174a113a66a27a112a170a147 and a174a74a145a33a112a170a147, on the same test set. |
| N03-1030 417 181:187 Another interesting nding is that the performance of current state-of-the-art syntactic parsers (Charniak, 2000) is not a bottleneck for coming up with a good solution to the sentence-level discourse parsing problem. |
| N03-1030 418 146:187 The discourse parsing model uses syntactic trees produced by Charniaks parser (2000) and discourse segments produced by the algorithm described in Section 3. |
| N03-1030 419 59:187 Given a sentence a2a44a43a46a45a39a47a12a45a49a48a51a50a52a50a4a50a10a45a49a53a54a50a4a50a52a50a55a45a49a56, we rst nd the syntactic parse tree a7 of a2 . We used in our experiments both syntactic parse trees obtained using Charniaks parser (2000) and syntactic parse trees from the PennTree bank. |
| N03-1030 420 122:187 For this evaluation, we re-trained Charniaks parser (2000) such that the test sentences from the discourse corpus were not seen by the syntactic parser during training. |
| N03-1030 421 5:187 1 Introduction By exploiting information encoded in human-produced syntactic trees (Marcus et al. , 1993), research on probabilistic models of syntax has driven the performance of syntactic parsers to about 90% accuracy (Charniak, 2000; Collins, 2000). |
| J01-2004 422 144:462 The parsers with the highest published broad-coverage parsing accuracy, which include Charniak (1997, 2000), Collins (1997, 1999), and Ratnaparkhi (1997), all utilize simple and straightforward statistically based search heuristics, pruning the search-space quite dramatically. |
| J01-2004 423 333:462 The differences between a k-best and a beam-search parser (not to mention the use of dynamic programming) make a running time difference unsur17 Our score of 85.8 average labeled precision and recall for sentences less than or equal to 100 on Section 23 compares to: 86.7 in Charniak (1997), 86.9 in Ratnaparkhi (1997), 88.2 in Collins (1999), 89.6 in Charniak (2000), and 89.75 in Collins (2000). |
| W06-1636 424 11:210 Instead researchers condition parsing decisions on many other features, such as parent phrase-marker, and, famously, the lexical-head of the phrase (Magerman, 1995; Collins, 1996; Collins, 1997; Johnson, 1998; Charniak, 2000; Henderson, 2003; Klein and Manning, 2003; Matsuzaki et al. , 2005) (and others). |
| W06-1623 425 81:250 We automatically segment sentences into clauses using a robust statistical parser (Charniak, 2000). |
| I05-6006 426 144:165 Charniak (2000) uses WSJ as both training and testing data and it is reasonable to expect a fairly good overlap in terms of lexical co-occurrences and linguistic structures and hence good performance scores. |
| I05-6006 427 141:165 Charniak (2000) reports a maximum entropy inspired parser that scored 90.1% average precision/recall when trained and tested with sentences from the Wall Street Journal corpus (WSJ). |
| I05-2036 428 13:116 All documents are parsed using Eugene Charniaks maximum entropy inspired parser (Charniak, 2000). |
| W06-2607 429 37:203 2.1 Basic SRL approach The SRL approach that we adopt is based on the deep syntactic parse (Charniak, 2000) of the sentence that we intend to annotate semantically. |
| W06-3601 430 63:298 In fact, the recursive transfer step can be done by a a linear-time algorithm (see Section 5), and the parsing step is also fast with the modern Treebank parsers, for instance (Collins, 1999; Charniak, 2000). |
| P06-1023 431 172:187 recall f-score this paper 86.0 82.3 84.1 Campbell 85.2 81.7 83.4 Dienes & Dubey 86.5 72.9 79.1 Johnson 85 74 79 Table 5: Accuracy of empty category prediction on section 23 The good performance of our parser on the empty element recognition task is remarkable considering the fact that its performance on the labeled bracketing task is 3% lower than that of the Charniak (2000) parser used by Campbell (2004). |
| P06-1021 432 16:158 In this separate-processing approach, reparanda are located through a variety of acoustic, lexical or string-based techniques, then excised before submission to a parser (Stolcke and Shriberg, 1996; Heeman and Allen, 1999; Spilker et al. , 2000; Johnson and Charniak, 2004). |
| P06-1021 433 99:158 The third experiment measures the benefit from syntactic indicators alone in Charniaks lexicalized parser (Charniak, 2000). |