RCE-OIE: Open Information Extraction Using a Rule-Based Clause Extraction Engine for Semantic Applications

Open Information Extraction (OIE) is a process of extracting clauses present in the text. Extraction of clauses is useful for several applications. However, the existing OIE methods do not focus on the improvement of such applications. In this paper, we present a methodology for OIE using a rule-based clause extraction engine (RCE-OIE) by considering some aspects like handling of coordinating conjunctions, negations, and relative clauses for the improvement of semantic applications. We have evaluated RCE-OIE on OIE datasets to show that our clause extraction approach is domain-independent and comparable with the state-of-the-art OIE systems. Our RCE-OIE is capable of improving the performance of downstream applications. In particular, RCE-OIE significantly improves the performance of paraphrase identification on Microsoft Research corpus when compared with the existing OIE systems.