Zavrel, J., Daelemans, W., Veenstra, J.: Resolving pp attachment ambiguities with memory-based learning. In: Proceedings CoNNL, Madrid, Computational Linguistics, Tilburg University (1997) 136 -- 144  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... They were printed by the NEGRA annotation tool [5] Memory based learning has recently been applied to a variety of NLP classification tasks, including part of speech tagging, noun phrase chunking, grapheme phoneme conversion, word sense disambiguation, and pp attachment (see [9] 14] [15] for details) construct tree(chunk list, treebank) while (chunk list is not empty) do remove first chunk from chunk list process chunk(chunk, treebank) Figure 3: Pseudo code for tree construction, main routine. process chunk(chunk, treebank) words : string yield(chunk) tree : complete ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP attachment ambiguities with memory-based learning. In M. Ellison, editor, Proceedings of the Workshop on Computational Natural Language Learning (CoNLL'97), Madrid, 1997.

....subset of the stored examples. An input instance is a labeled example (x; l) where x 2 V = f1; 2; vg k is a representation of the instance as a list of v valued tokens, and l 2 f1; 1g is the label. For example, for the task of disambiguating prepositional phrase attachment (PPA) Zavrel et al. 1997 ] k = 4 and the instance is represented as ( Verb, Noun, PP, NN) label) a sequence of four part ofspeech tags extracted from the prepositional phrase in the input sentence, along with the correct attachment l 2 fVerb; Noung; e.g. eat, cake, with, fork) Verb) Typically, k would be a small ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving pp attachment ambiguities with memory based learning. In Computational Natural Language Learning, Madrid, Spain, July 1997.

....attachment of a PP in the sequence VP NP PP (VP = verb phrase, NP = noun phrase, PP = prepositional phrase) The data consists of fourtuples of words, extracted from the Wall Street Journal Treebank. From the original data set, used by (Ratnaparkhi et al. 1994) Collins and Brooks, 1995) and (Zavrel et al. 1997), Daelemans et al. 1999) took the train and test set together to form the particular data also used here. Table 2 lists the average (10 fold crossvalidation) accuracies, measured in percentages of correctly classified test instances, of ib1 ig, ripper, and rbm on these five tasks. The clearest ....

J. Zavrel, W. Daelemans, and J. Veenstra. 1997. Resolving PP attachment ambiguities with memorybased learning. In M. Ellison, editor, Proc. of the Workshop on Computational Language Learning (CoNLL'97), ACL, Madrid.

....(linear combination) and outputs 0 if the result is below the threshold, and 1 otherwise. Wrongly predicted training examples make the weights of the model change, in a multiplicative way, to better fit the training set. 9 [55, 14, 56, 15] PoS tagging [61, 60, 90] PP attachment disambiguation [246], shallow parsing [227] and smoothing of probability estimates [245] The work of other authors include applications to partial parsing (chunking) and context sensitive parsing [210, 7, 33] WSD [159, 157, 84, 73] text categorization [183, 238, 237, 239] semantic interpretation [38] machine ....

....the references corresponding to different levels (lexical, syntactic, semantic, etc. of language acquisition. 12 DTs HMMs ME IBL Clause Boudaries [181] Shallow Parsing [45, 1, 16, 212] 211] 7, 227, 33, 58] Parsing [12, 132, 92] 178] 210, 37, 36, 38] PP attachment disambiguation [180] [246] TBL NB NNs LSM EC Clause Boudaries [95] Shallow Parsing [18] 128, 129] 130] 223] Parsing [115, 43] 93] PP attachment disambiguation [20] 52] 125, 218] 107] 2] Table 2: References corresponding to syntactic analysis and structural ambiguity NLP problems DLs DTs NB TBL EM WSD ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP attachment Ambiguities with Memory--Based Learning. In Proceedings of the Conference on Computational Natural Language Learning, CoNLL97, pages 136--144, Madrid, Spain, 1997.

....attachment of a PP in the sequence VP NP PP (VP = verb phrase, NP = noun phrase, PP = prepositional phrase) The data consists of fourtuples of words, extracted from the Wall Street Journal Treebank. From the original data set, used by (Ratnaparkhi et al. 1994) Collins and Brooks, 1995) and (Zavrel et al. 1997), Daelemans et al. 1999) took the train and test set together to form the particular data also used here. Table 2 lists the average (10 fold crossvalidation) accuracies, measured in percentages of correctly classi ed test instances, of ib1 ig, ripper, and rbm on these ve tasks. The clearest ....

J. Zavrel, W. Daelemans, and J. Veenstra. 1997. Resolving PP attachment ambiguities with memorybased learning. In M. Ellison, editor, Proc. of the Workshop on Computational Language Learning (CoNLL'97), ACL, Madrid.

.... These trees have proved to reduce significantly the space requirements and to be very efficient and accurate in several domains, including: Phonology (stress, word pronunciation) and morphology [Dae95, BDW96, DBG96, BWD98] POS tagging [DZBG96, DZB96, HZD98] PP attachment disambiguation [ZDV97], shallow parsing [Vee98] and smoothing of probability estimates [ZD97] The work of other authors include applications to partial parsing (chunking) and context sensitive parsing [SY92, ADK98] WSD [NL96, Ng97, FITT98] text categorization [RL94, YC94] semantic interpretation [Car94] machine ....

....1. References corresponding to some low level NLP tasks DTs HMMs ME IBL Clause Boudaries [RR97] Shallow Parsing [Chu88, Abn91, BSK97, SB98c] SB98b] ADK98, Vee98] Parsing [BJL 92a, Mag95a, HSO98] Rat97a] SY92, Car93b, Car93a, Car94] PP attachment disambiguation [RRR94] [ZDV97] TBL NB NNs LSM Clause Boudaries [HP94] Shallow Parsing [Bri93] Lyo94, LD95] Parsing [Leh91, CSL93] PP attachment disambiguation [BR94] CB95] L op98, SLL98] KR98] Table 2. References corresponding to syntactic analysis and structural ambiguity NLP problems 44 2. STATE OF THE ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP attachment Ambiguities with Memory--Based Learning. In Proceedings of the Conference on Computational Natural Language Learning, CoNLL97, pages 136--144, Madrid, Spain, 1997.

....of a PP in the sequence VP NP PP (VP = verb phrase, NP = noun phrase, PP = prepositional phrase) The data consists of four tuples of words, extracted from the Wall Street Journal Treebank. From the original data set, used by Ratnaparkhi, Reynar, and Roukos (1994) Collins and Brooks (1995) and Zavrel, Daelemans, and Veenstra (1997), Daelemans, Van den Bosch, and Zavrel (1999) took the train and test set together to form the particular data also used here. Part of speech tagging (henceforth pos) is the disambiguation of syntactic classes of words in particular contexts. We assume a tagger architecture that processes a ....

Zavrel, J., W. Daelemans, and J. Veenstra. 1997. Resolving PP attachment ambiguities with memory-based learning.

....selected subset of the stored examples. An input instance is a labeled example (x; l) where x 2 V = f1; 2; vg k is a representation of the instance as a list of v valued tokens, and l 2 f1; 1g is the label. For example, for the task of disambiguating prepositional phrase attachment (PPA) [17], k = 4 and the instance is represented as ( Verb, Noun, PP, NN) label) a sequence of four part ofspeech tags extracted from the prepositional phrase in the input sentence, along with the correct attachment l 2 fVerb; Noung; e.g. eat, cake, with, fork) Verb) Typically, k would be a small ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving pp attachment ambiguities with memory based learning. In Computational Natural Language Learning, Madrid, Spain, July 1997.

....verb. A contrasting sentence would be he eats pizza with anchovies : eats, pizza, with, anchovies, noun. From the original data set, used in statistical disambiguation methods by Ratnaparkhi, Reynar, and Roukos (1994) and Collins and Brooks (1995) and in a memory based learning experiment by Zavrel, Daelemans, and Veenstra (1997), Daelemans, Van den Bosch, and Zavrel, 1998 forthcoming) took the train and test set together to form the data also used here. For each task fambl is compared with ib1, ib1 ig, and igtree. Table 5 lists the generalisation accuracies obtained in these comparisons, on the four tasks. The results ....

Zavrel, J., W. Daelemans, and J. Veenstra. 1997. Resolving pp attachment ambiguities with memory-based learning. In M. Ellison, editor, Proc. of the Workshop on Computational Language Learning (CoNLL'97), ACL, Madrid.

.... for lexical tagging and partial parsing: e.g. part of speech tagging (Cardie 1993a, Daelemans et al. 1996) word sense disambiguation (Ng and Lee 1996) semantic class tagging (Cardie 1993a) noun phrase chunking (Argamon et al. 1998, Daelemans et al. 1999) and prepositional phrase attachment (Zavrel et al. 1997). Finally, case based methods have also been employed in higher level NLP tasks like context sensitive parsing (Simmons and Yu 1992) text categorization (Riloff and Lehnert 1994) and concept extraction (Cardie 1993a) It is well known in the machine learning community, however, that the success ....

J. Zavrel, W. Daelemans, and J. Veenstra. 1997. Resolving pp attachment ambiguities with memory-based learning. In Ellison, M., editor, Proceedings of the Workshop on Computational Language Learning (CoNLL '97). Association for Computational Linguistics.

....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 been made as to their applicativity to others. In this paper we cast the disambiguation problem as a learning problem and use tools from computational learning ....

Zavrel, J.; Daelemans, W.; and Veenstra, J. 1997. Resolving pp attachment ambiguities with memory based learning.

....pass (though separating subject verb detection from object verb detection is surely an option that must be investigated) In current research we are extending the approach to other types of constituent chunks and other types of syntactic relations. Combined with previous results on PP attachment [Zavrel et al. 1997], the results presented here will be integrated into a complete shallow parser. Acknowledgements This research was carried out in the context of the In duction of Linguistic Knowledge (ILK) research programme, supported partially by the Foundation of Language, Speech and Knowledge (TSL) ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving pp attachment ambiguities with memory-based learning. In M. Ellison, editor, Proc. o[ the Workshop on Computational Language Learning (CoNLL '97), ACL, Madrid, 1997. 6O

....have been fixed. The API has been adapted a bit to allow more practical use of it. 3.3 From version 1.0 to 2.0 We have added a new algorithm: TRIBL, a hybrid between the fast IGTREE algorithm and real nearest neighbor search (for more details, see 5. 4, or Daelemans, van den Bosch, and Zavrel (1997)) This algorithm is invoked with the a 2 switch and requires the specification of a so called TRIBL offset, the feature where IGTREE stops and case bases are stored under the leaves of the constructed tree. Support for numeric features. Although the package has retained its focus on discrete ....

....nearest neighbor position. For example, if the nearest neighbor is at a distance of one mismatch from the test instance, then the nearest neighbor set will contain the entire partition of the training set that matches all the other features but contains any value for the mismatching feature (see Zavrel and Daelemans (1997) for a more detailed discussion) With the MVDM metric, however, the nearest neighbor set will either contain patterns which have the value with the lowest (V 1 ; V 2 ) in the mismatching position, or MVDM will select a totally different nearest neighbor which has less exactly matching features, ....

[Article contains additional citation context not shown here]

Zavrel, J., W. Daelemans, and J. Veenstra. 1997. Resolving PP attachment ambiguities with memory-based learning. In M. Ellison, editor, Proc. of the Workshop on Computational Language Learning (CoNLL'97), ACL, Madrid.

....the attachment of a PP in the sequence VP NP PP (VP = verb phrase, NP = noun phrase, PP = prepositional phrase) The data consists of four tuples of words, extracted from the Wall Street Journal Treebank. From the original data set, used by (Ratnaparkhi et al. 1994) Collins Brooks, 1995) and (Zavrel et al. 1997), Daelemans et al. 1999) took the train and test set together to form the particular data also used here. More details on numbers of features, values per features, number of classes and number of instances are displayed in Table 1. The numbers of values for some features in the np and pp tasks ....

....distances are summed over all classes (and normalized) and are strictly speaking not feature weights, but using VDM does lead to the situation in which with k = 1 smaller numbers of nearest neighbors are used for classification. With larger k, generalization accuracies with VDM tend to increase (Zavrel Daelemans, 1997). Therefore, we see as relevant future research the investigation of larger k values in the kernel k NN classifier when unpacking multi valued features and classes. Acknowledgements The authors wish to thank Walter Daelemans, the anonymous reviewers, and the ILK (Tilburg) and CNTS (Antwerp) ....

Zavrel, J., Daelemans, W., & Veenstra, J. (1997). Resolving PP attachment ambiguities with memory-based learning. Proc. of the Workshop on Computational Language Learning (CoNLL'97). New Brunswick, NJ: ACL.

....model of parsing. Mbl models have been applied to several areas in language technology and human language processing tasks, such as word sense disambiguation (Veenstra, van den Bosch, Buchholz, Daelemans Zavrel, 2000) shallow parsing (Buchholz, Daelemans Veenstra, 1999) PP attachment (Zavrel, Daelemans Veenstra 1997), POS tagging (Daelemans, Zavrel, Berck Gillis, 1996) morphology (van den Bosch Daelemans 1999) and pronunciation (van den Bosch 1997) All these mbl applications are real life, performed on large amounts of data and show state of the art performance. Mbl stores instances in an instance base. ....

Zavrel, J., Daelemans, W., & Veenstra, J. (1997). Resolving PP Attachment Ambiguities with MemoryBased Learning In: Proceedings of the Conference on Computational Natural Language Learning (CoNLL97) . Madrid, Spain.

.... to a large range of Natural Language Processing tasks in our group: hyphenation and syllabification ( 17] assignment of word stress ( 11] grapheme to phoneme conversion ( 13] diminutive formation ( 16] morphological analysis ( 29, 28] part of speech tagging ( 18, 35] PP attachment ([36]) word sense disambiguation( 30] subcategorization ( 4] chunking (partial parsing) 31] and shallow parsing ( 10, 5] Relations to statistical language processing are discussed in [34] A partial overview paper is [9] The first dissertation length study devoted to the approach is [26] ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP attachment ambiguities with memory-based learning. In Mark Ellison, editor, Proc. of the Workshop on Computational Natural Language Learning (CoNLL'97), ACL, Madrid, 1997.

....pass (though separating subject verb detection from object verb detection is surely an option that must be investigated) In current research we are extending the approach to other types of constituent chunks and other types of syntactic relations. Combined with previous results on PP attachment [ Zavrel et al. 1997 ] the results presented here will be integrated into a complete shallow parser. Acknowledgements This research was carried out in the context of the Induction of Linguistic Knowledge (ILK) research programme, supported partially by the Foundation of Language, Speech and Knowledge (TSL) ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving pp attachment ambiguities with memory-based learning. In M. Ellison, editor, Proc. of the Workshop on Computational Language Learning (CoNLL'97), ACL, Madrid, 1997.

....use of the most similar instantiated schema or bucket for extrapolation. In statistical language modeling this is known as backed off estimation (Collins and Brooks, 1995; Katz, 1987) The distance metric defines a specific to general ordering (X OE Y : read X is more specific than Y , see also Zavrel and Daelemans (1997)) where the most specific schema is the schema with zero mismatches (i.e. an identical instance in memory) and the most general schema has a mismatch on every feature, which corresponds to the entire memory being retrieved. If information gain weights are used in combination with the overlap ....

....research. More recently, the approach has been applied to part of speech tagging (morphosyntactic disambiguation) morphological analysis, and the resolution of structural ambiguity (prepositional phrase attachment) Daelemans and Van den Bosch, 1996; Van den Bosch, Daelemans, and Weijters, 1996; Zavrel, Daelemans, and Veenstra, 1997). Whenever these studies involve a comparison of memory based learning to more eager methods, a clear advantage of memory based learning is reported. Cardie (1993; 1994) suggests a memory based learning approach for both (morpho) syntactic and semantic disambiguation and shows excellent results ....

[Article contains additional citation context not shown here]

Zavrel, J., W. Daelemans, and J. Veenstra. 1997. Resolving pp attachment ambiguities with memory-based learning. In M. Ellison, editor, Proc. of the Workshop on Computational Language Learning (CoNLL'97), ACL, Madrid.

.... been successfully applied to a large range of Natural Language Processing tasks: hyphenation and syllabification ( 15] assignment of word stress ( 9] grapheme to phoneme conversion ( 11] diminutive formation ( 14] morphological analysis ( 26] part of speech tagging ( 16] PP attachment ([31]) word sense disambiguation( 27] subcategorization ( 4] and chunking (partial parsing) 28] Relations to statistical language processing are discussed in [30] A partial overview paper is [8] The first dissertation length study devoted to the approach is [24] in which the approach is ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP attachment ambiguities with memory-based learning. In Mark Ellison, editor, Proc. of the Workshop on Computational Natural Language Learning (CoNLL'97), ACL, Madrid, 1997.

.... been successfully applied to a large range of Natural Language Processing tasks: hyphenation and syllabification ( 8] assignment of word stress ( 9] graphemeto phoneme conversion ( 11] diminutive formation ( 15] morphological analysis ( 25] part of speech tagging ( 12] PP attachment ([28]) Not yet published experimental results exist for word sense disambiguation, subcategorisation, and chunking (partial parsing) Relations to statistical language processing are discussed in [27] A partial overview paper is [7] The first dissertation length study devoted to the approach is ....

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP-attachment ambiguities with memory-based learning. In Mark Ellison, editor, Proc. of the Workshop on Computational Natural Language Learning (CoNLL'97), ACL, Madrid, 1997.

No context found.

Zavrel, J., Daelemans, W., Veenstra, J.: Resolving pp attachment ambiguities with memory-based learning. In: Proceedings CoNNL, Madrid, Computational Linguistics, Tilburg University (1997) 136 -- 144

No context found.

Jakub Zavrel, Walter Daelemans, and Jorn Veenstra. Resolving pp attachment ambiguities with memory-based learning. In Proceedings CoNNL, Madrid, pages 136 -- 144. Computational Linguistics, Tilburg University, 1997. 138

No context found.

Jakub Zavrel, Walter Daelemans, and Jorn Veenstra. Resolving pp attachment ambiguities with memory-based learning. In Proceedings CoNNL, Madrid, pages 136 -- 144. Computational Linguistics, Tilburg University, 1997. 138

No context found.

J. Zavrel, W. Daelemans, and J. Veenstra. Resolving PP attachment Ambiguities with Memory--Based Learning. In Proceedings of the Conference on Computational Natural Language Learning, CoNLL97, pages 136--144, Madrid, Spain, 1997.

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