도서관 챗봇 서비스의 질의응답 정합성 개선방법

Abstract

As non-face-to-face services became more important after the COVID-19 incident, the introduction of chatbots became essential for facilities that meet and use in face-to-face situations such as libraries. However, there are many cases where the compatibility of the response of the library chatbot is low, such as the difficulty of introducing the library chatbot, the error of the answer to the question of the chatbot user, and the recognition of the typo due to the lack of existing research on the library chatbot. In this study, we use the chatbot builder to provide a chatbot development guide and the chatbot builder to improve the response consistency of library users. In addition, we aim to develop a chatbot that provides easy-to-access service to users by linking chatbots and SNS, and provides query response services suitable for users' needs through button conversation and chatting. In this paper, we designed Q&A information through the data that summarizes and summarizes user requirements for services frequently visited by users through the chatbot menu composition of the National Assembly Library and the frequent question of the National Library. Based on the designed question and answer data, a scenario model was designed to enable users to query through button query and chat. The designed scenario is to build chatbots with improved response matching using the functions of chatbot builders: entity, API interlocking, and conversation flow. In order to evaluate the improvement of Q&A consistency in this study, the comparison analysis of the proposed method with the existing library chatbot, the National Assembly Library chatbot, was found to improve the Q&A consistency of users by the proposed method.