A Relevance feedback based approach for mixed script transliterated text search: Shared Task report by BIT Mesra, India

This paper describes the experiments carried out as part of the participation in FIRE-2014 Transliterated Search Shared task. We participated in subtask-2 and submitted two results generated by systems based on relevant feedback approach. Given a collection of documents in mixed script, the task is to retrieve relevant documents using queries in either script. The spelling variation between different versions of transliterated text results in a query mismatch problem which makes this task quite challenging. The basic idea behind our approach is the observation that small words posses very little spelling variation while transliterating it into a non native script. We proposed two n-gram approaches for query expansion based on number of characters in n-gram and their frequent occurrence. Using relevant feedback we got 14% gain in NDCG@1 and an improvement of 10% in MRR. The results suggest that our approach is quite helpful in retrieving relevant documents from mixed script document collection.