(A) computational analysis of linguistic activities of bilingual editors in wikipedia

Abstract

Language reveals a lot of information about its speakers. Speakers of one language usually share common cultural habits or regional characteristics, and their similarities become more obvious within the context where there are multiple languages in use. This thesis focuses on studying bilingual users of Wikipedia, one of the largest multilingual user-generated content platforms. Bilingual users tend to be active and make much contribution online. Among several online platform, Wikipedia is the largest multilingual platform where users from different linguistic background come together to express their thoughts and interests in English and other languages. To understand the specific topics edited by bilingual users, content analyses on Wikipedia user activities are performed by the use of topic models and other statistical measures. This thesis shows that bilinguals make different contribution from monolingual users. They tend to show and express interest in topics with locality, and there is a close association between language use and contribution. Bilingual users express their identity in terms of edition of national, cultural, and regional topics. However, the strengths of the association are different across language groups. The ratio of language use was another factor associated with Wikipedia contribution. This observation contributes to better understanding of bilingual users, and further provides a basis for advanced topic models.