Reciprocal Recommender Systems: Analysis of State-of-Art Literature, Challenges and Opportunities towards Social Recommendation

There exist situations of decision-making under information overload in the\nInternet, where people have an overwhelming number of available options to\nchoose from, e.g. products to buy in an e-commerce site, or restaurants to\nvisit in a large city. Recommender systems arose as a data-driven personalized\ndecision support tool to assist users in these situations: they are able to\nprocess user-related data, filtering and recommending items based on the users\npreferences, needs and/or behaviour. Unlike most conventional recommender\napproaches where items are inanimate entities recommended to the users and\nsuccess is solely determined upon the end users reaction to the\nrecommendation(s) received, in a Reciprocal Recommender System (RRS) users\nbecome the item being recommended to other users. Hence, both the end user and\nthe user being recommended should accept the 'matching' recommendation to yield\na successful RRS performance. The operation of an RRS entails not only\npredicting accurate preference estimates upon user interaction data as\nclassical recommenders do, but also calculating mutual compatibility between\n(pairs of) users, typically by applying fusion processes on unilateral\nuser-to-user preference information. This paper presents a snapshot-style\nanalysis of the extant literature that summarizes the state-of-the-art RRS\nresearch to date, focusing on the algorithms, fusion processes and fundamental\ncharacteristics of RRS, both inherited from conventional user-to-item\nrecommendation models and those inherent to this emerging family of approaches.\nRepresentative RRS models are likewise highlighted. Following this, we discuss\nthe challenges and opportunities for future research on RRSs, with special\nfocus on (i) fusion strategies to account for reciprocity and (ii) emerging\napplication domains related to social recommendation.\n