QeCSO: Design of hybrid Cuckoo Search based Query expansion model for efficient information retrieval

Abstract

The web contains lots of information that gets updated every second. Searching for a relevant document from the web needs an efficient scrutinization. As the user’s need varies based on location, intention and purpose the retrieval of an efficient response is a challenge. To address this challenge an information retrieval technique has been put forth along with the advent of the machine learning and deep learning models. We have proposed a QeCSO algorithm to perform an efficient retrieval of relevant response. The Attention-based Bi-directional LSTM (ATT-BLSTM) helps to improve the retrieval of the relevant document based on its feature that correlates the semantics between the query and the content. On further expanding the query we can observe a steep improvement in retrieving the response. To perform this, the output from ATT-BLSTM is given as an input to the meta-heuristic algorithm called cuckoo search. It helps us to retrieve the exact term to expand the query and make our search to move closer to an optimal solution. The performance of our approach is compared to those of other models based on the evaluation metrics such as accuracy and F -measure. It is evaluated by applying the model over the SQuAD 1.1 dataset. By analyzing the results it is verified that our proposed algorithm achieves an accuracy of 95.8% for an efficient information retrieval of relevant response with an increase in F -measure.