Title: Extraction of polarity comments from Nico Nico Douga

Authors: Hiroki Sakaji; Junya Ishibuchi; Hiroyuki Sakai

Addresses: Seikei University, 3-3-1 Kichijoji-Kitamachi, Musashino-shi, Tokyo, 180-8633, Japan ' Seikei University, 3-3-1 Kichijoji-Kitamachi, Musashino-shi, Tokyo, 180-8633, Japan ' Seikei University, 3-3-1 Kichijoji-Kitamachi, Musashino-shi, Tokyo, 180-8633, Japan

Abstract: In this paper, we propose two methods that extract polarity comments from Nico Nico Douga and analyse errors in these methods. For example, a positive comment is 'iine: good' and a negative comment is 'hetakuso: unskilled'. Online videos have become indispensable to people, with millions watching videos on video websites. Moreover, demand of video sites is increasing because internet election campaigns started in April 2013. We consider that polarity comments are useful for the analysis of internet election campaigns. Our methods achieved 99% and 65% precision for extracting positive and negative comments, respectively.

Keywords: text mining; information extraction; sentiment analysis; polarity comments; Nico Nico Douga; video sharing websites; online videos; internet election campaigns; data mining.

DOI: 10.1504/IJSSC.2016.080283

International Journal of Space-Based and Situated Computing, 2016 Vol.6 No.3, pp.165 - 172

Received: 28 Nov 2015
Accepted: 29 May 2016

Published online: 08 Nov 2016 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article