Good quality complementary information for multilingual wikipedia

  • Authors:

  • Yu Suzuki;Yuya Fujiwara;Yukio Konishi;Akiyo Nadamoto

  • Affiliations:

  • Nagoya University, Nagoya, Aichi, Japan;Konan University, Kobe, Hyogo, Japan;Konan University, Kobe, Hyogo, Japan;Konan University, Kobe, Hyogo, Japan

  • Venue:
  • WISE'12 Proceedings of the 13th international conference on Web Information Systems Engineering
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many Wikipedia articles lack information, because not all users submit truly complete information to Wikipedia. However, Wikipedia has many language versions that have been developed independently. Therefore, if we supply these complementary information from many language versions, the users must satisfy the amount of information of Wikipedia articles with the complementary information, instead of only one language version of Wikipedia articles. In this study, we specifically examine multilingual Wikipedia and propose a method of extracting good quality complementary information from Wikipedia of other languages. Specifically, we compare Wikipedia articles with less information to those with more information. From Wikipedia articles, which can have the same theme and different languages, we extract different information as complementary information. As described herein, we extract comparison target articles of Wikipedia based on a link graph, because cases exist in which information included in an articles is written in multiple pages of different languages. Furthermore, some low-quality information is extracted as complementary information because Wikipedia articles are written by not only good editors but also bad editors such as vandals. We propose a method to calculate the quality of information based on the editors, and we extract good quality complementary information.