2020 Volume 29 Issue 1 Pages 39-54
These days, in many companies, QA (Question Answering) systems accepting the query about the system and service online are introduced. In this QA system, a Web form and emails are adopted, and it is possible to send the query in an electronic document format. At this time, it is the present conditions that the answer document corresponding to the query document is made in hands. Therefore, it is considered useful to improve the efficiency of business operations in a company if automation of answer documents writing work is possible by utilizing past query and answer history data. However, the contents of query documents vary widely, such as cases where automatic response are possible, and cases where individual correspondence are required. Therefore, it is difficult to learn all the query and answer documents from the beginning and automate the writing of answer documents. Therefore, in this research, we build the model to analyze the relation between query and answer documents. Finally, we apply the proposed model to the actual data of the query and answer documents stored in the QA system of a certain major company and verify the effectiveness.