Home // eKNOW 2021, The Thirteenth International Conference on Information, Process, and Knowledge Management // View article

Extraction of Causal Relationships across Multiple Sentences from Securities Reports

Authors:
Takerou Aniya
Minoru Sasaki

Keywords: causal relationship extraction,securities report, pattern recognition

Abstract:
One of the most important sources of investment decisions in the stock market is the textual data contained in securities reports by companies. Investors consider investment strategies based on the information. However, since these text data are updated and published every day, it takes a great deal of time and money to read through and obtain information from all of them. In this study, we devised a method for extracting causal expressions from multiple sentences in securities reports. We extracted candidate causal expressions using clue expressions, and trained SVM (Support Vector Machine) by combining the similarity with the previous sentence and common phrases with the features obtained from a single sentence and verified how effective the method is. The effectiveness of the newly added features of inter-sentence similarity and common usage was confirmed.

Pages: 51 to 56

Copyright: Copyright (c) IARIA, 2021

Publication date: July 18, 2021

Published in: conference

ISSN: 2308-4375

ISBN: 978-1-61208-874-7

Location: Nice, France

Dates: from July 18, 2021 to July 22, 2021