An Empirical Study on Rule Granularity and Unification Interleaving in Unification-Based Parsers

概要

論文の詳細を見る
This paper describes an empirical study on the optimal granularity of the phrase structure rules and the optimal strategy for interleaving CFG parsing with unification in order to implement an efficient unification-based parsing system. We claim that using "medium-grained" CFG phrase structure rules balances the computational cost of CFG parsing and unification, and is a cost-effective solution for making unification-based grammar both efficient and easy to maintain. We also claim that "late unification, "which dalays unification until a complete CFG parse is found, reduces unnecessary copying of DAGs for irrelevant subparses and so improves performance significantly. The effectiveness of these methods is proven by an extensive experiment. The results show that, on average, the proposed system parses 2.1 times faster than our previous one. The grammar and the parser described in this paper have been fully implemented and used as the Japanese analysis module in SL-TRANS, the speech-to-speech translation system of ATR.
一般社団法人情報処理学会の論文
1994-05-15