/ プログラム/ 発表一覧/ 著者一覧/ 企業展示一覧/ jsai2012ホーム /

4M1-IOS-3c-1 Thai Wikipedia Quality Measurement using Fuzzy Logic

*セッションの無断動画配信はご遠慮下さい。

Tweet #jsai2012 このエントリーをはてなブックマークに追加

06月15日(Fri) 09:00〜12:20 M会場(-山口県自治会館/大会議室(80))
4M1-IOS-3c International Organized Session「Special Session on Web Intelligence & Data Mining (3)」

演題番号4M1-IOS-3c-1
題目Thai Wikipedia Quality Measurement using Fuzzy Logic
著者Kanjana Saengthongpattana(Kasetsart University, Bangkok, Thailand)
Nuanwan Soonthornphisaj(Kasetsart University, Bangkok, Thailand)
時間06月15日(Fri) 09:00〜09:30
概要Wikipedia is widely known as an online encyclopedia, one of success is collaboratively written and maintained articles by volunteers online. The open access model that is key to Wikipedia's success, however, can also be a source of the quality problems and not uniformly good quality. The Thai Wikipedia is assigning quality level including Featured Articles and Good Articles to its articles. As of February 2012, only 92 out of a total of 71,500 articles on the Thai Wikipedia are slated to be the featured articles and the good articles. From this number it's important to consider that why the good article have a tiny number and unreasonable to simply assume that the Thai Wikipedia is a completely reliable or unreliable. However, the degree of the article's quality is ambiguous due to some conditions can not be clearly determined in examples such as advertisement's style writing, missing neutrality, and less content. This characteristics are difficult to judge the quality although is done by professional. Many current techniques for automated method rely on exactly quality feature and identify quality article by classification or clustering method. Indeed, judge the quality of the article do not explicitly. Therefore we propose to use fuzzy logic as an approach to evaluate and predict the degree of the quality of the Thai Wikipedia articles. In this research, we have used a dataset which consist of 92 Thai Wikipedia qualified articles. Our evaluation is based on a corpus comprising of human labeled the degree of each quality this articles. We found that the degree of quality articles obtained from Fuzzy Logic provide the accuracy close to the expert inspection.
論文PDFファイル