A Multi-stage Chinese Collocation Extraction System
Most of the existing collocation extraction systems are based on globally significant statistical behaviors without mechanisms to handle different types of collocations. By taking compositionality, substitutability, modifiability and internal associations into consideration, collocations are categorized into four different types in this work. Based on the analysis for each type of collocation, a multi-stage extraction system is designed using different combinations of discriminative features so as to identify different types of collocations in different stages. Perceptron training is employed to optimize the consolidation of discriminative features from different sources. Experiment results show that the achieved performance is much better than most reported work.
KeywordsDiscriminative Feature Word Combination Context Window Dependency Grammar Collocation Type
Unable to display preview. Download preview PDF.
- 2.Smadja, F.: Retrieving Collocations from Text: Xtract. Computational Linguistics 19(1), 143–177 (1993)Google Scholar
- 3.Church, K., Hanks, P.: Word Association Norms, Mutual Information, and Lexicography. Computational Linguistics 16(1), 22–29 (1990)Google Scholar
- 4.Xu, R.F., Lu, Q., Li, Y.: An Automatic Chinese Collocation Extraction Algorithm Based on Lexical Statistics. In: Proc. IEEE Int. Conf. NLPKE (2003)Google Scholar
- 5.Lin, D.K.: Extracting Collocations from Text Corpora. In: First Workshop on Computational Terminology (1998)Google Scholar
- 6.Li, W.Y., Lu, Q., Xu, R.F.: Similarity based Chinese Synonyms Collocation Extraction. Int. J. Computational Linguistics and Chinese Language Processing 10(1), 123–144 (2005)Google Scholar
- 7.Mei, J.J.: Dictionary of Modern Chinese Collocations. Hanyu Dictionary Press (1999)Google Scholar
- 8.Robinson, J.J.: Dependency Structures and Transformation Rules. Language 46 (1970)Google Scholar
- 10.Mei, J.J., Zhu, Y.M.: Tong_Yi_Ci_Lin, Shanghai. Dictionary Press (1996)Google Scholar