Negotiation framework for group recommendation based on fuzzy computational model of trust and distrust
- PDF / 2,273,887 Bytes
- 28 Pages / 439.37 x 666.142 pts Page_size
- 55 Downloads / 250 Views
Negotiation framework for group recommendation based on fuzzy computational model of trust and distrust Nirmal Choudhary 1 & Sonajharia Minz 1 & K. K. Bharadwaj 1 Received: 18 August 2019 / Revised: 30 June 2020 / Accepted: 13 July 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Group recommender system (GRS) is the gradually prospering type of recommender system (RS) which tends to provide recommendations for the group of users rather than the individual. Most of the existing GRS obtain group preferences using equal weighing of the individual preferences, ignoring the relationship among group members within the group. But this is not a practical scenario because each member has different behavior. Therefore, in this article, we introduce a multiagent based negotiation mechanism between agents, each of them acts in favor of one group member. The proposed negotiation protocol allows agents to accept or discard a part of the offer based on trust and distrust among users, which gives more agility to the negotiation process. Further, we use memory for each agent in the group that records the previously proposed offers for that agent. The efficiency of trust-distrust enhanced GRSs is compared with traditional techniques and the outcomes of computational experiments confirm the supremacy of our proposed models over baseline GRSs techniques. Keywords Recommender systems . Group recommender systems . Multi-agent negotiation . Fuzzy trust and distrust . Memory
1 Introduction In the era of overwhelming growth and participation of billions of users on the Web, people depend upon many online applications to make decisions about which book to read, where to * Nirmal Choudhary [email protected] Sonajharia Minz [email protected] K. K. Bharadwaj [email protected]
1
School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India
Multimedia Tools and Applications
go in holidays, what to buy, and even what to eat. Recommender systems (RSs) help online users by providing the most useful, personal, and effective recommendations while using large information spaces [1, 3, 10, 34]. They suggest items related to their implicit or explicit preferences. Most of the traditional recommendation system techniques focused only on a single-user recommendation. But in everyday life, various activities that can be done with groups like watching a movie [31], going to a restaurant, listening to music [28], or traveling with friends [5, 14, 35], etc. This situation led to the development of group recommender systems (GRSs). The purpose of GRSs is to select a set of items that are likely of interest to a group of users [13, 15, 26]. The traditional group recommendation techniques employ aggregation strategies to generate group profiles. Though these techniques suffer from several shortcomings. First, aggregation strategies may not satisfy all group members in a unified manner. For example, we presume that three users rate an item with ratings 1, 2 and, 5, respectively. T
Data Loading...
