D. Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....have been applied to learn classi ers from corpora in order to perform WSD. These algorithms extract certain features from the annotated corpus and use them to form a representation of each of the senses. This representation can then be applied to new instances in order to disambiguate them [11 13]. In the framework of corpus based approaches, successful corpus based approaches to POS tagging which used Hidden Markov Models (HMM) have been extended in order to be applied to WSD. In [14] they estimated a bigram model of ambiguity classes from the SemCor corpus for the task of ....

Yarowsky, D.: Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In: Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM, ACL (1994) 88-95

....the part of speech of the word, and the syllable length of the word . As classes of the SVM learning, we use whether each hypothesized word is correct or incorrect. Since We compared the performance of SVM learning with much simpler machine learning techniques such as decision list learning [13], and found that SVM learning outperforms decision list learning. It is guaranteed that the two halves do not share speakers. We used : http: www.cs.cornell.edu People tj svm light ) as a tool for SVM learning. We also evaluated the effect of acoustic and language ....

D. Yarowsky, "Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French," in Proc. 32nd ACL, 1994, pp. 88--95.

....Confusion Set Disambiguation Several methods have been presented for confusion set disambiguation. The more recent set of techniques includes multiplicative weight update algorithms [4] latent semantic analysis [7] transformation based learning [8] differential grammars [10] decision lists [12], and a variety of Bayesian classifiers [2,3,5] In all of these papers, the problem is formulated as follows: Given a specific confusion set (e.g. to, two, too ) all occurrences of confusion set members in the test set are replaced by some marker. Then everywhere the system sees this marker, it ....

....as collocations surrounding the ambiguity site; these are essentially the same features as those used for the other disambiguation in stringcontext problems. 2.2 Learning Curves for NLP A number of learning curve studies have been carried out for different natural language tasks. Ratnaparkhi [12] shows a learning curve for maximum entropy parsing, for up to roughly one million words of training data; performance appears to be asymptoting when most of the training set is used. Henderson [6] showed similar results across a collection of parsers. Figure 1 shows a learning curve we generated ....

Yarowsky, D. (1994). Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proc. 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM.

.... d satisfying: arg max d#D P (d e 1 , e n ) arg max d#D P (d)P (e 1 , e n d) P (e 1 , e n ) P (e 1 , e n i=1 P (e i Another way of estimating the conditional probability distribution is to represent it in the form of a probabilistic decision list , as is proposed in (Yarowsky, 1994). Since a decision list is a sequence of IF THEN type rules, the use of it in disambiguation turns out to utilize only the strongest pieces of evidence. Yarowsky has also devised a heuristic method for e#cient learning of a probabilistic decision list. The merits of this method are ease of ....

Yarowsky, David. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95.

....Experimental results show that our approach is viable and we identify extensions for further improving performance. This work extends the statistical analysis of lexical cooccurrence to the statistical analysis of entire clauses. Lexical cooccurrences have been used for disambiguation problems [23, 14], and for the automatic identification of semantically related groups of words [19, 8] Our work differs in that cooccurrence is counted among similar clauses as opposed to individual words, we use cooccurrence of clausal features (e.g. tense) and a symbolic expression created automatically by ....

....to increase the performance of aspectual classification. Previous efforts in corpus based natural language classification have incorporated machine learning methods to coordinate multiple indicators, e.g. to classify adjectives according to markedness [9] and to perform accent restoration [23]. Klavans and Chodorow [12] describe why a weighted sum of multiple aspectual indicators could be advantageous. Aspectual Telic Non Telic Marker: Description: Frequency Frequency NotProgressive Not progressive tense 96.56 95.35 SpecialPerfect Perfect and not progressive 6.17 4.66 allMatch ....

D. Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in spanish and french. In Proceedings of the 32nd Annual Meeting of the ACL, San Francisco, CA, June 1994. Morgan Kaufmann.

....correct sense. The sense labels are typically taken from a dictionary. Given this information, a supervised learning algorithm constructs rules that achieve high discrimination between occurrences of different word senses. Examples of supervised learning methods for WSD appear in [1] 6] 9] [22], 18] The learning methods used in those studies are general purpose, including: decisiontree induction, decision list induction, feed forward neural networks with backpropagation and nave Bayesian learning. Their results are very encouraging, exceeding 90 correct sense labelling in some ....

....use a larger context window ( 50 words on each side) None of the fairly recent approaches presented above uses purely local information. Yarowsky [21] and Schtze [17] present purely topical methods, but in both papers the value of local information is noted. Most of the recent approaches, e.g. [22], 18] combine local and topical information, in order to improve their performance. A critical component of any application of machine learning is the representation of the training examples and the generated model, i.e. the disambiguator here. The most popular representation for training ....

Yarowsky, D.: Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the Annual Meeting of the Association for Computational Linguistics, (1994) 88-95

....relations (4d) 21] The results for both are given separately in Table 1. The only global feature in this experiment is a bag of word for the words in the sentence (knowledge types 4b, 4c) 14] We chose to use one of the simplest yet effective way to combine the features: decision lists [22]. The decision list orders the features according to their log likelihood, and the first feature that is applicable to the test occurrence yields the chosen sense. In order to use all the available data, we used 10 fold cross validation. Table 1 shows the results in the 8 noun set for each of the ....

Yarowsky, D.: Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. Proceedings of the ACL (1994)

....if requires, as well as its adaptability to a new domain or a new de nition of named entities. In general, creating training data for supervised learning is somewhat easier than creating pattern matching rules by hand. Next, we apply Yarowsky s method for supervised decision list learning 1 (Yarowsky, 1994) to 1 We choose the decision list learning method as the Table 1: Statistics of NE Types of IREX frequency ( NE Type Training Test ORGANIZATION 3676 (19.7) 361 (23.9) PERSON 3840 (20.6) 338 (22.4) LOCATION 5463 (29.2) 413 (27.4) ARTIFACT 747 (4.0) 48 (3.2) DATE 3567 (19.1) 260 (17.2) ....

....i M NE m( 3) M R 1 M R 2 (2) Current Position) 4 Supervised Learning for Japanese Named Entity Recognition This section describes how to apply the decision list learning method to chunking tagging named entities. 4. 1 Decision List Learning A decision list (Rivest, 1987; Yarowsky, 1994) is a sorted list of decision rules, each of which decides the value of a decision D given some evidence E. Each decision rule in a decision list is sorted in descending order with respect to some preference value, and rules with higher preference values are applied rst when applying the ....

[Article contains additional citation context not shown here]

D. Yarowsky. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proc. of the 32nd Annual Meeting of ACL, pages 88-95.

....selected ambiguous words, or artificial pseudo words. 7 The senses of the example have been taken from the WordNet 1. 5 [146] 8 These examples have been taken from the DSO corpus [159] 15 Many standard ML algorithms for supervised learning have been applied to WSD, including: Decision Lists [242], Neural Networks [224] Bayesian learning [27] Exemplar Based learning [156, 84] and Boosting [72] Further, in [150] some of the previously cited methods are compared, jointly with Decision Trees and Rule Induction algorithms, on a very restricted domain. The performance of supervised ML ....

D. Yarowsky. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95, Las Cruces, NM, 1994. ACL.

....because they avoid to some extent the data fragmentation problem. Thus, regarding NLP, they have been applied to lexical ambiguity resolution. In particular: Word sense disambiguation, lexical choice in machine translation, homograph disambiguation in speech synthesis and accent restoration [240, 243, 241, 150, 208]. 2.2.3 Transformation Based Error Driven Learning (TBL) TBL was introduced by Brill in the early 90s, as a new approach to corpus based natural language learning. The learning algorithm is a mistake driven greedy procedure that produces a set of rules. It works iteratively by adding at each ....

.... TBL NB Acquisition of verbal properties [221, 209] General machine translation [10] 100] Spelling correction [133] 86, 88, 89] DLs ILP NNs Clust GAs LSM LogL Acquisition of verbal properties [152, 153, 29] 209] 209, 135] General machine translation [235] Spelling correction [241, 208] [116] 89] Generation [176] Table 5: References corresponding to Machine Translation and other NLP tasks 14 3 Word Sense Disambiguation: A Case Study in Supervised Machine Learning The present section is devoted to explain the comparison between four machine learning algorithms applied to ....

D. Yarowsky. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95, Las Cruces, NM, 1994. ACL.

....diacritics can be seen as a special case of spelling correction. In particular, the same division on word based methods and methods using the context is present. The methods that use the context are similar to those used in spelling correction. For a description of those methods see [Yar94a] and [Yar94b] The word based methods for the restoration of diacritics are also the basis for the context based methods by providing them with choices. The techniques used in them are partially dioeerent from those used in spelling correction. In particular, if the word from a text is present in the ....

David Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in spanish and french. In Proceeding of the ACL'94, San Francisco, California, 1994. Association for Computational Linguistics, Morgan Kaufmann.

....ers from corpora in order to perform WSD. Generally, supervised approaches 1 have obtained better results than unsupervised methods on small sets of selected ambiguous words, or arti cial pseudo words. Many standard ML algorithms for supervised learning have been applied, such as: Decision Lists (Yarowsky, 1994; Agirre and Martinez, 2000) Neural Networks (Towell and Voorhees, 1998) Bayesian learning (Bruce and Wiebe, 1999) Exemplar Based learning (Ng, 1997a; Fujii et al. 1998) Boosting (Escudero et al. 2000a) etc. Unfortunately, there have been very few direct comparisons between alternative ....

D. Yarowsky. 1994. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88{ 95, Las Cruces, NM. ACL.

....have obtained better results than unsupervised methods on small sets of selected highly ambiguous words, or artificial pseudo words. Many standard ML algorithms for supervised learning have been applied, such as: Bayesian learning [16, 19] Exemplar based learning [18, 16, 5] Decision Lists [21], Neural Networks [20] etc. Further, Mooney [15] provides a comparative experiment on a very 1 TALP Research Center, Software Department, Technical University of Catalonia, Jordi Girona Salgado 1 3, Barcelona E 08034, Catalonia, email: fescudero, lluism, g.rigaug lsi.upc.es restricted domain ....

D. Yarowsky, `Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French', in Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics. ACL, (1994).

....because they avoid to some extent the data fragmentation problem. Thus, regarding NLP, they have been applied to lexical ambiguity resolution. In particular: Word sense disambiguation, lexical choice in machine translation, homograph disambiguation in speech synthesis and accent restoration [Yar93, Yar94b, Yar94a, Moo96]. 3.2.3. Transformation Based Error Driven Learning (TBL) TBL was introduced by Brill in the early 90s, as a new approach to corpus based natural language learning. The learning algorithm is a mistake driven greedy procedure that produces a set of rules. It works iteratively by adding at ....

.... [Tan96, Sie97] General machine translation [BPP96] Jon96] Spelling correction [MB97] GCY93, Gol95, GR98] DLs ILP NNs Clust GAs LSM LogL Acquisition of verbal properties [MC95, MC96, CM97] Sie97] Sie97, MLC98] General machine translation [YPM96] Spelling correction [Yar94a] [Lew98] GR98] Table 5. References corresponding to Machine Translation and other NLP tasks small surveys. They are roughly included in the following list: WRS96, Car96b, DZBG96, Die97, SCVS98b] 4. A Machine learning Oriented Review of Decision Trees Decision trees are a way to represent ....

D. Yarowsky. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95, Las Cruces, NM, 1994. ACL.

.... the Catalan Research Department (CIRIT s consolidated research group 1999SGR 150, CREL s Catalan WordNet project and CIRIT s grant 1999FI 00773) Many standard ML algorithms for supervised learning have been applied, such as: Naive Bayes [19, 22] 19, 10] Exemplar based learning Decision Lists [28], Neural Networks [27] etc. Further, Mooney [17] has also compared all previously cited methods on a very restricted domain and including Decision Trees and Rule Induction algorithms. Unfortunately, there have been very few direct comparisons of alternative methods on identical test data. ....

Yarowsky, D.: Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd annual Meeting of the Association for Computational Linguistics, ACL, 1994.

....et al. 1993) and the DSO corpus (Ng and Lee, 1996) which have been tagged with word senses from WordNet. Besides we test an algorithm that automatically acquires training examples from the Web (Mihalcea Moldovan, 1999) In this paper we focus on one of the most successful algorithms to date (Yarowsky 1994), as attested in the Senseval competition (Kilgarriff Palmer, 2000) We will evaluate it on both SemCor and DSO corpora, and will try to test how far could we go with such big corpora. Besides, the usefulness of hand tagging using WordNet senses will be tested, training on one corpus and testing ....

....design of the experiments. The experiments are organized in three sections: experiments on SemCor and DSO, cross corpora experiments, and tagging SemCor using the Web data for training. Finally some conclusions are drawn. 1 Decision lists and the features used Decision lists (DL) as defined in (Yarowsky, 1994) are simple means to solve ambiguity problems. They have been successfully applied to accent restoration, word sense disambiguation and homograph disambiguation (Yarowsky, 1994; 1995; 1996) It was one of the most successful systems on the Senseval word sense disambiguation competition (Kilgarriff ....

[Article contains additional citation context not shown here]

Yarowsky, D. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French', in Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pp. 88--95. 1994.

....(those that learn from previously semantically annotated corpus) have obtained better results than unsupervised methods on small sets of selected ambiguous words, or artificial pseudo words. Many standard ML algorithms for supervised learning have been applied to WSD, including: Decision Lists (Yarowsky, 1994), Neural Networks (Towell and Voorhees, 1998) Bayesian learning (Bruce and Wiebe, 1999) Exemplar based learning (Ng, 1997a; Fujii et al. 1998) and Boosting (Escudero et al. 2000a) Further, in (Mooney, 1996) some of the previously cited methods are compared, jointly with Decision Trees and ....

D. Yarowsky. 1994. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, ACL, pages 88--95, Las Cruces, NM, USA.

....have obtained better results than unsupervised methods on small sets of selected highly ambiguous words, or artificial pseudo words. Many standard ML algorithms for supervised learning have been applied, such as: Bayesian learning [15, 18] Exemplar based learning [17, 15, 5] Decision Lists [20], Neural Networks [19] etc. 1 TALP Research Center, Software Department, Technical University of Catalonia, Jordi Girona Salgado 1 3, Barcelona E 08034, Catalonia, email: fescudero, lluism, g.rigaug lsi.upc.es Further, Mooney [14] provides a comparative experiment on a very restricted domain ....

D. Yarowsky, `Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French', in Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pp. 88--95, Las Cruces, NM, (1994). ACL. 6

.... probabilities of n grams type features (in the word or POS space) Machine learning based classi ers and maximum entropy models which, in principle, are not restricted to features of these forms have used them nevertheless, perhaps under the in uence of probabilistic methods (Brill, 1995; Yarowsky, 1994; Ratnaparkhi et al. 1994) It has been argued that the information available in the local context of each word should be augmented by global sentence information and even information external to the sentence in order to learn better classi ers and language models. E orts in this directions ....

D. Yarowsky. 1994. Decision lists for lexical ambiguity resolution: application to accent restoration in Spanish and French. In Proc. of the Annual Meeting of the ACL, pages 88-95.

....on both the issues of named entity chunking and classification in Japanese named entity recognition, and evaluate named entity chunking techniques in the context of minimally supervised learning. First, as a supervised learning method, we employ the supervised decision list learning method of (Yarowsky, 1994), into which we incorporate several noun phrase chunking techniques (sections 3. and 4. We chose the decision list learning method as the supervised learning technique because it is easy to implement and quite straightforward to extend a supervised learning version to a minimally supervised ....

....which we incorporate several noun phrase chunking techniques (sections 3. and 4. We chose the decision list learning method as the supervised learning technique because it is easy to implement and quite straightforward to extend a supervised learning version to a minimally supervised version (Yarowsky, 1994; Yarowsky, 1995; Collins and Singer, 1999) section 5. Then, we applied a minimally supervised learning algorithm to Japanese named entity recognition, where a list of frequent named entities are extracted from unlabeled data by a human and fed to the learning algorithm as seeds. The minimally ....

[Article contains additional citation context not shown here]

Yarowsky, D., 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proc. 32nd ACL.

....corpus) have obtained better results than unsupervised methods on small sets of selected highly ambiguous words, or artificial pseudo words. Many standard ML algorithms for supervised learning have been applied, such as: Naive Bayes [21, 23] Exemplar based learning [20, 21, 10] Decision Lists [31], Neural Networks [28] etc. Further, Mooney [19] provides a comparative experiment on a very restricted domain between all previously cited methods but also including Decision Trees and Rule Induction algorithms. Unfortunately, there have been very few direct comparisons of alternative methods on ....

Yarowsky, D.: Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd annual Meeting of the Association for Computational Linguistics, ACL, 1994.

.... occurring only when w 1 and w 2 are equal) Define f d (w 1 #w 2 ) as the number of co occurrences of w 1 , w 2 (in this order) in a sentence within a distance of no more than d words (weused = 3) Define the context of the occurrence of a word as the 1 local windowof Sigmad words. Yarowsky (Yarowsky, 1994) and Leacocketal. Leacock et al. 1998) suggest that this window size is optimal. With N denoting the length of the corpus (in words) Nd approximizes the total number of cooccurrence pairs. So, the estimated probabilityofapair P (w 1 #w 2 ) is: P (w 1 #w 2 ) f d (w 1 #w 2 ) Nd The ....

Yarowsky, D. (1994). Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the ACL, pages 88--95, Las Cruces, New Mexico.

.... or POS space) Machine learning based classi ers and maximum entropy models which, in principle, are not restricted to features of these forms have used them nevertheless, perhaps under the in uence of probabilistic methods, although in some cases have used features from both sides (Brill, 1995; Yarowsky, 1994; Ratnaparkhi et al. 1994) E orts in this directions consists of (1) directly adding syntactic information, as in (Chelba and Jelinek, 1998; Rosenfeld, 1996) and (2) indirectly adding it, via similarity models that; 3 in these case n gram type features are used whenever possible, and when ....

D. Yarowsky. 1994. Decision lists for lexical ambiguity resolution: application to accent restoration in Spanish and French. In Proc. of the Annual Meeting of the ACL, pages 88-95.

.... required by the manual approach and its inherent inflexibility led to the pursuit of ML techniques for the automatic induction of disambiguation rules (Hindle, 1989; Brill, 1995) or equivalent inference devices such as decision trees (Schmid, 1994b; Daelemans et al. 1996) or decision lists (Yarowsky, 1994). The accuracy of rule tree based taggers is comparable to that of stochastic taggers, yet they are much faster. Moreover, rules or decision trees lists are human understandable, thus it can be verified whether or not they capture true underlying linguistic phenomena. The bulk of the literature ....

Yarowsky, D. (1994) Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. Proceedings of ACL '94.

....of WSD on a predefined set of words in preannotated corpora. Some recent work on WSD includes the use of the knowledge contained in a machinereadable dictionary (Luk 1995) Wilks et al. 1990) supervised learning from tagged sentences (Bruce Wiebe 1994) Miller 1990) Ng Lee 1996) Yarowsky 1992)(Yarowsky 1994), unsupervised learning from raw Copyright c fl1999, American Association for Artificial Intelligence (www.aaai.org) All rights reserved. corpora (Yarowsky 1995) and hybrid methods that combine several knowledge sources, collocation and others (Bruce Wiebe 1994) Ng Lee 1996) Despite the ....

Yarowsky, D. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics.

....The knowledge based approach carries out disambiguation by using information from a lexicon or a knowledge base. The lexicon may be a machine readable dictionary, thesaurus or may be hand crafted rules, such as WordNet [2] 86] 104] 72] 63] LDOCE [46] 29] and Roget s International Thesaurus [111]. Wilk and Stevenson [109] Harley and Glennon [48] and McRoy [69] all use large lexicons and the information associated with the senses such as part of speech tags, topical guides and selectional preferences to indicate the correct sense. Yarowsky [111] measures the similarity of words by ....

....[29] and Roget s International Thesaurus [111] Wilk and Stevenson [109] Harley and Glennon [48] and McRoy [69] all use large lexicons and the information associated with the senses such as part of speech tags, topical guides and selectional preferences to indicate the correct sense. Yarowsky [111] measures the similarity of words by treating text as an unordered bag of words and looking at the semantic similarity as measured from the knowledge source between all the words in a certain window. The corpus based approach is also called the supervised learning approach. This approach applies ....

[Article contains additional citation context not shown here]

D. Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in spanish and french. Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, 1994.

....the basis of the immediate context and a set of events observed previously (the training corpus) Note, however, that this idea is not entirely original. El B ze et al. 3] describe an accent restoration technique that draws upon the same concepts, while Yarowsky has obtained comparable results [6] by combining different criteria for statistical disambiguation within a unifying framework (decision lists) 2 Automatic Accent Restoration We have developed an automatic accent restoration program called Reacc. It is based on a stochastic language model. Reacc will accept as input a string of ....

Yarowsky, David. 1994. "Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French." In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics (ACL--94), pp. 88--95.

....acute. In addition, huge word trigram tables need to be available at run time. Moreover, word trigrams are ineffective at capturing longdistance properties such as discourse topic and tense. Feature based approaches, such as Bayesian classifiers (Gale, Church, and Yarowsky, 1993) decision lists (Yarowsky, 1994), and Bayesian hybrids (Golding, 1995) have had varying degrees of success for the problem of context sensitive spelling correction. However, we report experiments that show that these methods are of limited effectiveness for cases such as ftheir ; there; they 0 reg and fthan; theng, where the ....

....the same part of speech. In this case, a more effective approach is to learn features that characterize the different contexts in which each word tends to occur. A number of feature based methods have been proposed, including Bayesian classifiers (Gale, Church, and Yarowsky, 1993) decision lists (Yarowsky, 1994), Bayesian hybrids (Golding, 1995) and, more recently, a method based on the Winnow multiplicative weight updating algorithm (Golding and Roth, 1996) We adopt the Bayesian hybrid method, which we will call Bayes, having experimented with each of the methods and found Bayes to be among the ....

Yarowsky, David. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95, Las Cruces, NM.

....within about the last half dozen years. A number of methods have been proposed, either for context sensitive spelling correction directly, or for related lexical disambiguation tasks. The methods include word trigrams (Mays et al. 1991) Bayesian classifiers (Gale et al. 1993) decision lists (Yarowsky, 1994), Bayesian hybrids (Golding, 1995) a combination of part of speech trigrams and Bayesian hybrids (Golding and Schabes, 1996) and, more recently, transformation based learning (Mangu and Brill, 1997) latent semantic analysis (Jones and Martin, 1997) and differential grammars (Powers, 1997) ....

.... . They therefore capture aspects of the context with a wide scope, semantic 5 flavor, such as discourse topic and tense. Collocations, in contrast, capture the local syntax around the target word. Similar combinations of features have been used in related tasks, such as accent restoration (Yarowsky, 1994) and word sense disambiguation (Ng and Lee, 1996) We use a feature extractor to convert from the initial text representation of a sentence to a list of active features. The feature extractor has a preprocessing phase in which it learns a set of features for the task. Thereafter, it can convert a ....

Yarowsky, D. (1994). Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proc. 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM.

....collocations with content words than those with function words. 4 In general, the high reliability of this behavior (in excess of 97 for adjacent content words, for example) makes it an extremely useful property for sense disambiguation. A supervised algorithm based on this property is given in (Yarowsky, 1994). Using a decision list control structure based on (Rivest, 1987) this algorithm integrates a wide diversity of potential evidence sources (lemmas, inflected forms, parts of speech and arbitrary word classes) in a wide diversity of positional relationships (including local and distant ....

....life are delicately discovered at a St. Louis plant manufacturing computer manufacturing plant and adjacent . the proliferation of plant and animal life . 6 Including variants of the EM algorithm (Baum, 1972; Dempster et al. 1977) especially as applied in Gale, Church and Yarowsky (1994). 7 Indeed, any supervised classification algorithm that returns probabilities with its classifications may potentially be used here. These include Bayesian classifiers (Mosteller and Wallace, 1964) and some implementations of neural nets, but not Brill rules (Brill, 1993) STEP 2: For each ....

[Article contains additional citation context not shown here]

Yarowsky, David, "Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French," in Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM, 1994.

....X into tag Y . M = max X;Y 2X freq(X) freq(Y ) freq(X) freq(Y ) Delta incontext(Y; C) Gamma incontext(X; C) 4. 3 Logarithmic Following the same idea of comparing all pairs of tags in a particular tag set, we can use a logarithmic scoring function, similar to the approach of Yarowski [Yar95]. This measure compares the frequencies of the two tags in the current context. As in the previous two measures, the comparison is adjusted by the overall frequencies of the tags. M = max X;Y 2X freq(X) freq(Y ) log incontext(Y; C) freq(Y ) Delta freq(X) incontext(X; C) 4.4 ....

David Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in spanish and french. Proceedings of ACL, 1995.

....general class of programs, currently our only convincing experimental results are on the past tense problem. The decision list mechanism in general should be applicable to other language problems (as evidenced by the use of propositional decision lists for problems such as lexical disambiguation [24]. Many realistic problems consist of rules with exceptions, and experimental results on additional applications are needed to support the general utility of this representation. 7 Conclusions Learning the past tense of English is a small by interesting subproblem in language acquisition which ....

David Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95, Las Cruces, NM, 1994.

.... corpora in order to identify co occurrence patterns for words and their prosodic realization [SHY92, RO96] These are used in combination with a decision algorithm to identify the best match of the current text to the learned patterns and thereby, choose the most appropriate pronunciation [Yar94] or prosody [WH92, SHY92, HP94, PM98] The corpus based approaches add semantic and usage information to text to speech systems, which typically have none. While they do not directly provide semantics, they do identify patterns of use that implicitly express semantic relations and constraints. ....

....the many features of language, text and speech. The collection and ordering of these predicates attempts to capture the many ways that a target may prime for a stimulus, and the many feature combinatorics that contribute to retrieval. It is somewhat the inverse of Yarowsky et al. s [SHY92, Yar96, Yar94] decision list algorithm for identifying homographs. In their work, the test that yields the highest score determines the pronunciation of the word or phrase. It is a best match approach. For Loq comparisons, the particular features that match are not as important as whether the cue and target ....

David Yarowsky. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Conference of the Association for Computational Linguistics, page 8 Pages, 1994.

.... method for acquiring decision lists for lexical 5 Stative verbs are verbs whose actions are assumed to hold at all times after their assertion (e.g. know, believe, love) the actions of non stative verbs are not always assumed to hold after assertion (e.g. fix, walk) 21 ambiguity resolution [Yarowsky , 1994] . Yarowsky s decision lists are ordered lists of single antecedent rules each entry in the list tests one feature and produces a classification. Unlike Brill s transformation based rule set, only the first applicable pattern in the list is applied to a novel instance. The decision lists are ....

Yarowsky, David. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32th Annual Meeting of the ACL, 1994.

....inference. Developing learning techniques for language disambiguation has been an active field in recent years and a number of statistics based and machine learning techniques have been proposed. A partial list consists of Bayesian classifiers (Gale, Church, Yarowsky 1993) decision lists (Yarowsky 1994), Bayesian hybrids (Golding 1995) HMMs (Charniak 1993) inductive logic methods (Zelle Mooney 1996) memorybased methods (Zavrel, Daelemans, Veenstra 1997) and transformation based learning (Brill 1995) Most of these have been developed in the context of a specific task although claims have ....

Yarowsky, D. 1994. Decision lists for lexical ambiguity resolution: application to accent restoration in Spanish and French. In Proc. of the Annual Meeting of the ACL, 88--95.

....data is available. Keywords: word sense disambiguation, decision lists, supervised machine learning, lexical ambiguity resolution, senseval 1. Introduction Decision lists have been shown to be effective at a wide variety of lexical ambiguity resolution tasks including word sense disambiguation (Yarowsky, 1994, 1995; Mooney, 1996; Wilks and Stevenson, 1998) text to speech synthesis (Yarowsky, 1997) multilingual accent diacritic restoration (Yarowsky, 1994) named entity classification (Collins and Singer, 1999) and spelling correction (Golding, 1995) One advantage offered by interpolated decision ....

.... Decision lists have been shown to be effective at a wide variety of lexical ambiguity resolution tasks including word sense disambiguation (Yarowsky, 1994, 1995; Mooney, 1996; Wilks and Stevenson, 1998) text to speech synthesis (Yarowsky, 1997) multilingual accent diacritic restoration (Yarowsky, 1994), named entity classification (Collins and Singer, 1999) and spelling correction (Golding, 1995) One advantage offered by interpolated decision lists (Yarowsky, 1994, 1997) is that they avoid the training data fragmentation problems observed with decision trees or traditional non interpolated ....

[Article contains additional citation context not shown here]

Yarowsky, D.: 1994, Decision lists for lexical ambiguity resolution: application to accent restoration in Spanish and French. Proceedings of ACL '94, pp. 88--95.

No context found.

D. Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM, 1994.

No context found.

Yarowsky,David. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88#95, Las Cruces, NM.

No context found.

Yarowsky, D. (1994). Decision Lists for Lexical Ambiguity Resolution : Application to Accent Restoration in Spanish and French. Proc. of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95.

No context found.

David Yarowsky. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the ACL, pages 88--95, Las Cruces, New Mexico, June 1994. Association for Computational Linguistics. 32

No context found.

David Yarowsky. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95. Association for Computational Linguistics, Somerset, New Jersey, 1994.

No context found.

D. Yarowsky, "Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French" Proceedings of the ACL, 1994: 88-95

No context found.

David Yarowsky. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95.

No context found.

David Yarowsky. 1994. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of ACL-94.

No context found.

Yarowsky D.: Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32th Annual Meeting of the Association for Computational Linguistics, #ACL'94#. 1994.

No context found.

David Yarowsky. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Assocation for Computational Linguistics, pages 88 95.

No context found.

David Yarowsky. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Assocation for Computational Linguistics, pages 88 95.

No context found.

David Yarowsky. 1994. Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In 32th Annual Meeting of the Associtation of the Computational Linguistics, pages 88--95.

No context found.

D. Yarowsky. Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, pages 88--95, Las Cruces, NM, 1994. ACL.

No context found.

David Yarowsky. 1994b. Decision Lists for Lexical Ambiguity Resolution: Applications to Accent Restoration in Spanish and French. In Proceedings of ACL-94, Las Cruces, New Mexico.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC