Exploiting Chunk-level Features to Improve Phrase Chunking

Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 557–567, Jeju Island, Korea, 12–14 July 2012. c⃝2012 Association for Computational Linguistics Exploiting Chunk-level Features to Improve Phrase Chunking Junsheng Zhou Weiguang Qu Fen Zhang Jiangsu Research Center of Information Security & Privacy Technology School of Computer Science and Technology Nanjing Normal University. Nanjing, China, 210046 Email:{zhoujs,wgqu}@njnu.edu.cn zf9646@126.com Abstract Most existing systems solved the phrase chunking task with the sequence labeling approaches, in which the chunk candidates cannot be treated as a whole during parsing process so that the chunk-level features cannot be exploited in a natural way. In this paper, we formulate phrase chunking as a joint segmentation and labeling task. We propose an efficient dynamic programming algorithm with pruning for decoding, which allows the direct use of the features describing the internal characteristics of chunk and the features capturing the correlations between adjacent chunks. A relaxed, online maximum margin training algorithm is used for learning. Within this framework, we explored a variety of effective feature representations for Chinese phrase chunking. The experimental results show that the use of chunk-level features can lead to significant performance improvement, and that our approach achieves state-of-the-art performance. In particular, our approach is much better at recognizing long and complicated phrases. 1 Introduction Phrase chunking is a Natural Language Processing task that consists in dividing a text into syntactically correlated parts of words. Theses phrases are non-overlapping, i.e., a word can only be a member of one chunk (Abney, 1991). Generally speaking, there are two phrase chunking tasks, including text chunking (shallow parsing), and noun phrase (NP) chunking. Phrase chunking provides a key feature that helps on more elaborated NLP tasks such as parsing, semantic role tagging and information extraction. There is a wide range of research work on phrase chunking based on machine learning approaches. However, most of the previous work reduced phrase chunking to sequence labeling problems either by using the classification models, such as SVM (Kudo and Matsumoto, 2001), Winnow and voted-perceptrons (Zhang et al., 2002; Collins, 2002), or by using the sequence labeling models, such as Hidden Markov Models (HMMs) (Molina and Pla, 2002) and Conditional Random Fields (CRFs) (Sha and Pereira, 2003). When applying the sequence labeling approaches to phrase chunking, there exist two major problems. Firstly, these models cannot treat globally a sequence of continuous words as a chunk candidate, and thus cannot inspect the internal structure of the candidate, which is an important aspect of information in modeling phrase chunking. In particular, it makes impossible the use of local indicator function features of the type "the chunk consists of POS tag sequence p1...,pk". For example, the Chinese NP " 农 业 /NN(agriculture) 生 产 /NN(production) 和/CC(and) 农村/NN(rural) 经济 /NN(economic) 发 展 /NN(development)" seems relatively difficult to be correctly recognized by a sequence labeling approach due to its length. But if we can treat the sequence of words as a whole and describe the formation pattern of POS …

Content truncated to 500 words