ech T Press Science Computers, Materials & Continua DOI:10.32604/cmc.2022.021033 Article BERT for Conversational Question Answering Systems Using Semantic Similarity Estimation Abdulaziz Al-Besher 1 , Kailash Kumar 1 , M. Sangeetha 2, * and Tinashe Butsa 3 1 College of Computing and Informatics, Saudi Electronic University, Riyadh, 11673, Kingdom of Saudi Arabia 2 Department of Information Technology, SRM Institute of Science and Technology, Kattankulathur, India 3 Department of Information Technology, Harare Institute of Technology, Belvedere, Harare * Corresponding Author: M. Sangeetha. Email: sangeetk@srmist.edu.in Received: 20 June 2021; Accepted: 27 July 2021 Abstract: Most of the questions from users lack the context needed to thor- oughly understand the problem at hand, thus making the questions impossible to answer. Semantic Similarity Estimation is based on relating user’s questions to the context from previous Conversational Search Systems (CSS) to provide answers without requesting the user’s context. It imposes constraints on the time needed to produce an answer for the user. The proposed model enables the use of contextual data associated with previous Conversational Searches (CS). While receiving a question in a new conversational search, the model determines the question that refers to more past CS. The model then infers past contextual data related to the given question and predicts an answer based on the context inferred without engaging in multi-turn interactions or requesting additional data from the user for context. This model shows the ability to use the limited information in user queries for best context inferences based on Closed-Domain-based CS and Bidirectional Encoder Representations from Transformers for textual representations. Keywords: Semantic similarity estimation; conversational search; multi-turn interactions; context inference; BERT; user intent 1 Introduction Conversational search is one of the most critical areas in Natural Language Processing (NLP); hence, researchers’ ambition is to understand user intent in multi-turn conversations to simulate human-to-human interaction in Conversational Assistants (CA). CSS can be defned as an approach to fnd information in a multi-turn conversation, and it has long been associated with Information retrieval systems. The adoption of CA in Conversational Search Systems (CSS) is currently rising, which has attracted much attention from researchers. The most common framework for CA mainly focuses on Natural Language Understanding (NLU) [1] to design and develop systems that can better understand human language. The objective is to understand NLP and identify the informational users’ needs (user intent) from natural language by analysing textual information. This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.