Hierarchical Maximum Pattern Matching with Rule Induction Approach for Sentence Parsing
Yi-Syun Tan, Yuan‐Cheng Chu, Jui‐Feng Yeh · 2012
Chinese parsing has been a highly active re-search area in recent years. This paper de-scribes a hierarchical maximum pattern matching to integrate rule induction approach for sentence parsing on traditional Chinese parsing task. We have analyzed and extracted statistical POS (part-of-speech) tagging in-formation from training corpus, then used the related information for labeling unknown words in test data. Finally, the rule induction regulation was applied to extract of the struc-ture of short-term syntactic and then per-formed maximum pattern matching for long-term syntactic structure. On Sentence Parsing task, our system performs at 44 % precision, 53 % recall, and F1 is 48 % in the formal test-ing evaluation. The proposed method can achieve the significant performance in tradi-tional Chinese sentence parsing. 1