Framework of Judgment System for Smart Home Assistant Utilizing Collective Intelligence Case-Based Reasoning

Abstract

This paper proposes a framework of judgment system for smart home assistant that utilizes Collective Intelligence Case Based Reasoning (CI-CBR). CBR is suitable for the smart home environment with its system adaptability to the changeful user scenarios. However, existing CBR solutions have shown relatively low accuracy in service recommendation. This research therefore aims at enhancing the accuracy by introducing collective intelligence into the recommendation system. Assuming that multiple agents will make better decision than single agent, we adopted a multi-agent approach to generate the most similar case, which represents the optimal recommendation from the case base. This paper describes how our system enables agents adopting different similarity measures come to an agreement about the most similar case by the means of majority voting in the judging process. Our framework of a collective judgment system demonstrates its potentials to improve recommendation accuracy, and further enhance the performance of existing smart home assistants.