O uso de chatbots para auxiliar a recuperação da informação em repositórios institucionais

Abstract

The study aims to analyze, based on national and international literature, how NLP and chatbots can contribute to the retrieval of information in institutional repositories. Three specific objectives were established: identify and map the main problems in research on NLP and chatbots in the area of information retrieval; analyze the solutions proposed in each study, highlighting approaches to improve information retrieval; characterize the specific applications of each study in order to improve information retrieval in institutional repositories, highlighting the practical contributions and potential benefits to the information search and retrieval process. The research adopts the content analysis methodology. Thematic categorical analysis was used to analyze the results obtained, providing a deeper understanding of the trends and patterns identified. The analysis categories were established based on the four-dimensional space in which the concept of relevance can be formally defined, for the purposes of evaluating information retrieval systems. Therefore, the research was based on two categories: information resources and representation of the user's problem. The research analyzes 263 articles published between 2019 and 2023, selected through the Web of Science database. The results highlight the importance of using NLP and chatbots in optimizing information retrieval in institutional repositories. These technologies have the potential to transform the way users interact and access information, providing a more efficient, personalized and accessible search. From the results, it can be inferred that a task-driven repository and the construction of chatbots based on existing, intuitive and open source tools is a way of applying the research analyzed. It is concluded that there is a promising scenario in the use of NLP and chatbots to improve information retrieval in institutional repositories, simplifying the research process and making it more efficient for users. For future research, there are opportunities to explore advanced NLP techniques to improve chatbots' understanding of complex queries and personalize the user experience by adapting the interaction style and providing relevant content recommendations.