Improving the Multilingual User Experience of Wikipedia Using Cross-Language Name Search

Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the ACL, pages 492–500, Los Angeles, California, June 2010. c⃝2010 Association for Computational Linguistics Improving the Multilingual User Experience of Wikipedia Using Cross-Language Name Search Raghavendra Udupa Microsoft Research India Bangalore, India. Mitesh Khapra ∗ Indian Institute of Technology Bombay Powai, India. Abstract Although Wikipedia has emerged as a power- ful collaborative Encyclopedia on the Web, it is only partially multilingual as most of the content is in English and a small number of other languages. In real-life scenarios, non- English users in general and ESL/EFL 1 users in particular, have a need to search for rele- vant English Wikipedia articles as no relevant articles are available in their language. The multilingual experience of such users can be significantly improved if they could express their information need in their native language while searching for English Wikipedia arti- cles. In this paper, we propose a novel cross- language name search algorithm and employ it for searching English Wikipedia articles in a diverse set of languages including Hebrew, Hindi, Russian, Kannada, Bangla and Tamil. Our empirical study shows that the multilin- gual experience of users is significantly im- proved by our approach. 1 Introduction Since its inception in 2001, Wikipedia has emerged as the most famous free, web-based, collaborative, and multilingual encyclopedia with over 13 million articles in over 270 languages. However, Wikipedia exhibits severe asymmetry in the distribution of its content in the languages of the world with only a small number of languages dominating (see Table ∗This work was done when the author was a summer intern at Microsoft Research India. 1English as Second Language and English as Foreign Lan- guage. 1). As a consequence, most users of the under- represented languages of the world have no choice but to consult foreign language Wikipedia articles for satisfying their information needs. Table 1: Linguistic asymmetry of Wikipedia Language Speakers Contributors Articles English 1500M 47.1% 3,072,373 Russian 278M 5.2% 441,860 Hebrew 10M 0.7% 97,987 Hindi 550M 0.06% 50,926 Bangla 230M 0.02% 20,342 Tamil 66M 0.04% 19,472 Kannada 47M 0.02% 7,185 Although consulting foreign language Wikipedia is not a solution for the problem of linguistic asym- metry, in the specific case of ESL/EFL users who form a sizable fraction of Internet users of the world 2, it is arguably the most practical option today. Typ- ically, ESL/EFL users are reasonably good at read- ing and extracting relevant information from English content but not so good at expressing their infor- mation needs in English. In particular, getting the spellings of foreign names in English correctly is very difficult for most ESL/EFL users due to the dif- ferences in the way a foreign name is pronounced in the native languages. For instance, Japanese EFL speakers often break consonant clusters in for- eign names using vowels (see Table 2) and Hindi ESL speakers find it difficult to differentiate between ‘an’, ‘en’, and ‘on’ in English names (such as ‘Clin- 2As per some estimates, …

Content truncated to 500 words