Extraction of the Argument Structure of Tokyo Metropolitan Assembly Minutes: Segmentation of Question-and-Answer Sets

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 2064–2068 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 2064 Extraction of the Argument Structure of Tokyo Metropolitan Assembly Minutes: Segmentation of Question-and-Answer Sets Keiichi Takamaru1, Yasutomo Kimura2, Hideyuki Shibuki3, Hokuto Ototake4, Yuzu Uchida5, Kotaro Sakamoto6, Madoka Ishioroshi3, Teruko Mitamura7, Noriko Kando3,8 1Utsunomiya Kyowa University, Japan 2Otaru University of Commerce, Japan 3National Institute of Informatics, Japan 4Fukuoka University, Japan 5Hokkai-Gakuen University, Japan 6Yokohama National University, Japan 7Carnegie Mellon University, USA 8SOKENDAI, Japan takamaru@kyowa-u.ac.jp Abstract In this study, we construct a corpus of Japanese local assembly minutes. All speeches in an assembly were transcribed into a local assembly minutes based on the local autonomy law. Therefore, the local assembly minutes form an extremely large amount of text data. Our ultimate objectives were to summarize and present the arguments in the assemblies, and to use the minutes as primary information for arguments in local politics. To achieve this, we structured all statements in assembly minutes. We focused on the structure of the discussion, i.e., the extraction of question and answer pairs. We organized the shared task “QA Lab-PoliInfo” in NTCIR 14. We conducted a “segmentation task” to identify the scope of one question and answer in the minutes as a sub task of the shared task. For the segmentation task, 24 runs from five teams were submitted. Based on the obtained results, the best recall was 1.000, best precision was 0.940, and best F-measure was 0.895. Keywords: Argument mining, Shared task, Local assembly, Tokyo Metropolitan assembly 1. Introduction Numerous local autonomies in Japan provide open access to a variety of political documents via their websites. These documents include basic urban development plans, local assembly minutes, and ordinances. Such information ob- tained through the internet can be used to compare lo- cal autonomies and identify their individual characteris- tics. Local assembly minutes are crucial for determining such characteristics because they include various represen- tatives ’positions on the policies enforced by an autonomy. Some studies have been conducted by political scientists and econometricians to analyze local assembly minutes . (Kawaura et al., 2018). However, these studies have raised issues concerning the analysis methods of such minutes. One issue is concerned with the different ways in which these minutes are released to the public. There are 47 pre- fectures and several cities, towns, and villages in Japan; local assembly minutes are made available in a variety of ways. Gathering local assembly minutes and presenting the data collected in a unified format for analysis at a national level is therefore expensive. For this reason, in this study, we attempt to build the corpus of Japanese local assem- bly minutes. All speeches in an assembly are transcribed into a local assembly minutes based on the local autonomy law. Therefore, the local assembly minutes result in an ex- tremely large amount of text data. For example, the number of characters of the Tokyo Metropolitan Assembly minutes on February 26, 2013 …

Content truncated to 500 words