Non-factoid Question Answering System of Management Philosophy based on Lecture Transcripts and Writings of a Business Leader

Abstract

In recent years, companies that commit fraud or misconduct are increasing. It is thought that these things happen due to the lack of corporate rules and philosophy for business management. In this paper, we propose a question answering system of management philosophy using transcribed lecture and writings of Dr. Kazuo Inamori, who is one of the most respected business leaders in the world. In the proposed system, we first analyze the question sentence and use the query of nouns obtained by morphological analysis of the sentence and the expanded query obtained using word2vec, and search the documents using Okapi BM25. Then, we identify and extract the sentences that should be answers to the question by using a deep learning technique, and show the extracted sentences to the user in a ranked form.