ABSTRACT
Question answering (QA) systems have developed into useful resources for automatically providing relevant and precise responses to user queries. QA systems aim to understand the semantics and context of the questions and retrieve relevant information to provide accurate and meaningful answers. These systems play a crucial role in information retrieval, knowledge extraction, and human-computer interaction. A question-answering (QA) system built on a pipeline architecture is presented in this study, leveraging the power of transformer-based language models, with a focus on the Bidirectional Encoder Representations from Transformers (BERT) model. The proposed QA system aims to efficiently and accurately provide answers to user queries in natural language, addressing the challenges of understanding context and retrieving relevant information from large text corpora. The Stanford Question Answering Dataset (SQuAD) dataset was used in this experiment. In our trial, the passage retrieval accuracy applying BERT was 79.37%. This model's accuracy was measured as 60.23% against the Medical Question Answering Dataset (MedQuAD), a dataset from the medical field.
