Abstract
Recommender systems are able to suggest the most suitable items to a given user, taking into account the user’s and item`s data. Currently, these systems are offered almost everywhere in the online world, such as in e-commerce websites, newsletters, or video platforms. To improve recommendations, the user’s context should be considered to provide more accurate algorithms able to achieve higher payoffs. In this paper, we propose a pre-filtering recommendation system that considers the context of a coworking building and suggests the best workplaces to a user. A cyber-physical context-aware multi-agent system is used to monitor the building and feed the pre-filtering process using fuzzy logic. Recommendations are made by a multi-armed bandit algorithm, using -greedy and upper confidence bound methods. The paper presents the main results of simulations for one, two, three, and five years to illustrate the use of the proposed system.
Highlights
The recommendation of items to users in commerce is not something new and already existed long before the invention of computers
This paper proposes a novel context-aware recommender system using a pre-filtering process with fuzzy data collected by a context-aware multi-agent system (MAS) that monitors a coworking building
The use of a cyber-physical MAS allows real-time monitoring of context data inside the coworking building, to be later fuzzified and used in the pre-filtering process. This combination of MAS, fuzzy data, and pre-filtering enables the development of a novel context-aware recommender system that can recommend workplaces inside the coworking building
Summary
The recommendation of items to users in commerce is not something new and already existed long before the invention of computers. This paper proposes a novel context-aware recommender system using a pre-filtering process with fuzzy data collected by a context-aware multi-agent system (MAS) that monitors a coworking building. The use of a cyber-physical MAS allows real-time monitoring of context data inside the coworking building, to be later fuzzified and used in the pre-filtering process. This combination of MAS, fuzzy data, and pre-filtering enables the development of a novel context-aware recommender system that can recommend workplaces inside the coworking building. The proposed system overcomes existing systems by combining real-time data, fuzzification, pre-filtering, a multi-armed bandit algorithm, and user feedback.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
