Automatic Event Extraction method for Analyzing Text Narrative Structure

Hye-Yeon Yu, Moon-Hyun Kim

Abstract: This paper presents an analysis of contemporary methods for event extraction from text narratives and of various event expression formats. It also briefly discusses future directions in narrative understanding and generation using artificial intelligence. The three-step study method for extracting events from text stories, comprising token analysis and part-of-speech tagging, dependent parsing, and standardization work, is analyzed. Expressions created using a tuple format are compared and contrasted with expressions created using the 5W format. Finally, we propose a novel method to organize events in a tuple format, reconstructing compound and complex sentences as simple sentences. Our method identifies and extracts verbs, subject, object, and preposition phrases. It then automatically extracts the multiple events that comprise each sentence.