Neural Networks in Chinese Lexical Classification

Language, Information and Computation(PACLIC 11),1996, 119-128 Neural Networks in Chinese Lexical Classification Md Maruf Hasan Kim-Teng Lua Department of Information Systems & Computer Science National University of Singapore {maruf, luakt} @iscs.nus.sg Abstract Lexical attributes, like syntactic (part-of-speech) and semantic (semantic category) attributes, are in most cases, ambiguous in every languages. Automatic resolution of ambiguity of these attributes can be achieved using different techniques; rule-based, statistical, NN-based and their hybrids. Moreover, one linguistic feature also has influence over the resolution of ambiguity of another feature; eg.. knowledge of syntactical category can assist smooth disambiguation of semantic category and vice versa. Properly disambiguated syntactic and semantic properties of lexicon may significantly help us in word sense disambiguation, text analysis, information retreival, natural language understanding and speech processing etc. In this paper, we have presented our neural network based Classification Tool. We have used this tool in Part-of-Speech tagging and Semantic- Category tagging of Chinese lexicon with the help of thesaurus and large training corpus. Experimental results are analysed and compared. 1 Introduction to Lexical ambiguity in Chinese Lexical attributes, such as part-of-speech and semantic category disambiguation in Chinese is more or less similar to that of western languages. We explain an example of such ambiguity of the Chinese lexicon, q;:f.t (BuZhun). In sentence, itX *-* PAM [ZheLi BuZhun Xi Yan / Smoking is prohibited here], the part-of-speech and semantic category of the word, T* is verb and prohibition respectively. But in another sentence, [Wo De ShouBiao BuZhun / My watch is (showing) inaccurate (time)], the same word, *.i.t belongs to the category of adjective and similarities and differences respectively. Disambiguation of part-of-speech is well addressed by Greene et. al. (1973), Brill (1994) (rule-based: English); Chang et. al. (1993) (statistical: Chinese); Church (1988), Kempe (1993) (statistical: English, German); Nakamura et. al. (1990), Schmid (1994) (connectionist: English, German). To our knowledge, very few people has dealt with the connectionist approach for the Chinese lexicon. Word sense disambiguation is well addressed by Lam (1995) (dictionary-clue word: Chinese); Lesk et. al. (1986) (dictionary-clue word: English); DeRose (1988), Yarowsky (1992) (statistical: English); Ide et. al. 119 (1990), Cottrell (1989) (connectionist: English). Again, to our knowledge, connectionist approaches are less explored for Chinese lexical research. In disambiguation of Chinese lexicon, also there are approaches to combine word-selection (segmentation) and part-of-speech tagging together (Chang et. al. 1993, Bai 1995). In English NLP, Richard and Bronwyn (1995) used semantic categorical attributes to assist part-of-speech tagging too. Yarowsky (1992) has made use of part-of-speech tags information to assign semantic category tags in his statistical model of word sense disambiguation. He has calculated concordances of 100 surrounding words (50 to the left and 50 to the right) to decide the right category tag to the current word. Nakamura et. al. (1990), in their speech recognition system, trained a 4- layered feed-forward network with up to three preceding words' tags as input to predict the word category (POS) of the current word. Using this predictor, they were able to enhance their Speech Recognition …

Content truncated to 500 words