A CBR-based conversational architecture for situational data management

Abstract

This paper introduces a conversational Case-Based Reasoning (CBR) architecture, aimed at improving situational data management by incorporating user feedback into the process. The core of the architecture is a “human-in-the-loop” approach implemented through a conversational agent, which facilitates interaction between the user and the system. The CBR-based approach leverages a historical knowledge base that is dynamically updated based on user feedback, allowing for a more responsive and adaptive system. This feedback plays a crucial role in the processes of case retrieval, review, and retention within the CBR cycle, enabling the system to evolve based on user interactions. An empirical study involving 22 participants was conducted to assess the impact of user feedback on system recommendations. This study included both static and dynamic test scenarios, focusing on aspects such as visibility, support, usefulness, and data integration. The results highlighted a general preference for recommendations that were influenced by user input, indicating the effectiveness of incorporating human feedback in the decision-making process. The research contributes to situational data management by illustrating how a conversational CBR framework, integrated with user feedback, can improve processes such as data integration and data discovery. In addition, it highlights the importance of user involvement in enhancing the functionality of conversational systems for complex data management, pointing to the potential for further development in this area.