Home // eKNOW 2021, The Thirteenth International Conference on Information, Process, and Knowledge Management // View article
Authors:
Ryo Kato
Minoru Sasaki
Keywords: stance classification, Approval or disapproval forecast.
Abstract:
This paper uses the meeting minutes of the Tokyo Metropolitan Assembly and the Ibaraki Prefectural Assembly to create a rule base for predicting the approval or disapproval of each faction, and to examine the effectiveness of methods for extracting opinions from the meeting minutes and methods for estimating when opinions cannot be obtained. In recent years, the voter turnout in Japan has been on a downward trend. In order to solve this problem, we conducted this research because we believe that it is necessary to present materials to judge the credibility of politicians' statements by clearly indicating in an easy-to-understand manner what kind of opinion each faction has on each proposition and the direction of its argument. In the Tokyo Metropolitan Assembly, we used the meeting minutes of the plenary sessions, and from the statements made during the meetings, we assumed that the opinion of the faction to which the speaker belonged was the opinion of "for" or "against" the proposal. As a method of estimating opinions, we created and verified several methods of estimating opinions when they could not be read. In addition to the method used in the Tokyo Metropolitan Assembly, the Ibaraki Prefectural Assembly implemented an estimation method that applied the opinions of the Assembly Steering Committee and an estimation method that used machine learning. As a result, in the Tokyo Metropolitan Assembly, the method that applied the opinions of the minority group obtained the highest value. For the Ibaraki Prefectural Assembly, the method that applied the members of the Assembly Steering Committee obtained the highest value. For the estimation of the Tokyo Metropolitan Assembly's opinion, there are similar studies that have recorded higher values, and further improvement in accuracy can be expected by using machine learning.
Pages: 42 to 46
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