An Approach for Building Content Recommendation System for Bilinguals

Abstract

AbstractLanguage diversity in India is an extension of the vast cultural diversity that exists in Indian society. The globalization and presence of multinational companies made an inevitable need of English language, usage including in academics, though there are regional universities that use local languages as the medium of instruction. The emerging digital content market cannot elude the language diversity prevalent in the Indian population. The paper explores the possibility of finding common patterns among Hindi-speaking university-educated netizens regarding their first language preferences. These common patterns could be used to build a recommendation system for multilingual content-providing systems. The dataset that is used for building the system is collected as part of pilot work for creating a database in line with the Language Experience and Proficiency Questionnaire (LEAP-Q) for Indian languages. The questionnaire is formalized by the Language Technology Research Centre, Hyderabad, in collaboration with Pauranik Neuro Center, Indore. The questionnaire comprises questions related to language preferences and usage in different situations and reading habits. The questionnaire is devised to assess the Hindi language usage among bilingual Hindi native speakers. For this purpose, respondents are university-educated Native Hindi language users. The analysis shows that there is little relation to the language preferences between formal language usage and informal language usage. The reading habits of Hindi newspaper is observed as more of a personal choice, rather depends on any other language preferences. The analysis shows that three to four variables are enough to estimate preferences of the first language reading habit (Hindi). The proposed method could be used to recommend the right language content to bilingual readers at an average accuracy of around 50%.KeywordsDigital platformsBilingualsRecommendation system