Syntactic analysis of long sentences based on S-clauses

Abstract

In dependency parsing of long sentences with fewer subjects than predicates, it is difficult to recognize which predicate governs which subject. To handle such syntactic ambiguity between subjects and predicates, an "S(ubject)-clause" is defined as a group of words containing several predicates and their common subject, and then an automatic S-clause segmentation method is proposed using semantic features as well as morpheme features. We also propose a new dependency tree to reflect S-clauses. Trace information is used to indicate the omitted subject of each predicate. The S-clause information turned out to be very effective in analyzing long sentences, with an improved parsing performance of 4.5%. The precision in determining the governors of subjects in dependency parsing was improved by 32%.