The effectiveness of retrieving relevant information is often hindered by ambiguous and short queries, compounded by often imprecise initial translation in Cross Lingual Information Retrieval (CLIR). These limitations are still a challenge for Query Reformulation (QR) techniques, which primarily focus on selecting effective expansion terms but generally neglect the impact of determining their optimal placement within the query. This paper introduces an approach to QR that not only focuses on identifying contextually relevant expansion terms but also effectively determines their optimal placement within the query to maximize retrieval performance. The method integra Continuous Bag of Words (CBOW) and Term Frequency-Inverse Document Frequency (TF-IDF) techniques to extract meaningful expansion terms from a snippet dataset. Co-occurrence-based term placement algorithm has also been proposed to find the optimal location for term placement. Further, we observed the improvements of 23.42 and 17.39% in the retrieval effectiveness when the expansion term added at optimal location and manually at the end, which underline the importance of both precise term selection and optimal term positioning in improving CLIR effectiveness.
Asthana et al. (Sun,) studied this question.