Semantic desktop search application for Hindi-English code-mixed user query with query sequence analysis

With the advancement of technology, the hard disk space in our personal computers is getting bigger. Now a days it seems easy to search the web and get the relevant result than searching our personal computers. This is because of absence of any sort of interconnection between the documents, media files, and rich text present in the PC. It is really tough to create a knowledge graph for the same. In this rapidly growing digitization world, in county like India where we have listed 22 languages in the constitution the task gets even tough. People from different demographic and socio-economic background are expected to adapt to this fast evolving scenario. Under the influence of social media due deep penetration of smart phones and internet, Indian users are taking up to this. If we observe closely the users are more comfortable in typing their native language in Roman script (i.e. English.) In this work we have developed a desktop search application which can also take Hindi-English Code-Mixed user query. Subsequently the query is converted into correctly spelled English words. Then the query is analyzed for inherent sequential continuity and final reformulation is done. Code-Mixing query is frequently observed in social media and search engines especially form multilingual users. We have considered Hindi-English mixed term queries written in common roman script. Secondly, we have also included the query sequence analysis to automate the Query Reformulation process. The user may reformulate the query if he fails to get the relevant result. Here we expect the user to enter the query like text based conversational search system. On the backend, the data on the PC in indexed into the Apache Solr using Apache Tika and other dependency parsers.