Bidirectional Encoding Contextual Approach for Identification of Relevant Document in Corpus

Abstract

With the increasing advance of computer and information technologies, numerous documents have been published online as well as offline, and as new research fields have been continuingly created, users have a lot of trouble in finding their interesting documents. These documents can be in the form of blogs, research papers, and thesis. There is a heterogeneous set of documents which has information linked with each other. Traditional search is about taking an input of the query text from the user and checking if the subsequence is a part of any sentence in the set of documents and showing the set to the user. In this paper, we have proposed a Bidiection Encoding Contextual algorithm that can be applied to different types of documents and do a semantic search across the corpus. The algorithm used to understand the meaning of the word, their relative relationship between other words and provide the user with the documents that not just has the textual reference but also contain the relative meaning of the query. On the COVID-19 dataset, test been performed on the reliability of the interpretation through the function of linguistic similarities. The experimental findings demonstrate the strong association between the conceptual term interpretation of human consciousness in the role of measuring the similarity. Experiments show that the Bidirectional Encoding Contextual model has the best accuracy of 85.6% when compared with other traditional models like RNN, CNN and LSTM models.