Cross language information retrieval for code mixed Kannada English queries

Abstract

Though there has been a lot of Natural Language Processing work done towards languages, such as English, Arabic, Chinese similar efforts for Indian languages is still in its infancy. Kannada is one of popular south Indian languages. It is proposed in this research work to experiment Information Retrieval of Kannada English code mixed queries. In order to mitigate the aforementioned issue, a paradigm which understands and processes code mixed queries is proposed. The entire work is divided into four modules. Initially, Language Identification module identifies language of each word of the code mixed query which is followed by back transliterating Romanised kannada words in the query to its native script using transliteration module. The mixed script query resulted from previous module is fed to a translation model to obtain two monolingual queries of Kannada and English languages. Finally, monolingual search queries of Kannada and English languages are used to retrieve relevant documents for the input code mixed query. newline