A literature review of recommender systems in the television domain

Recommender Systems (RSs) are software tools and techniques providing suggestions of relevant items to users. These systems have received increasing attention from both academy and industry since the 1990s, due to a variety of practical applications as well as complex problems to solve. Since then, the number of research papers published has increased significantly in many application domains (books, documents, images, movies, music, shopping, TV programs, and others). One of these domains has our attention in this paper due to the massive proliferation of televisions (TVs) with computational and network capabilities and due to the large amount of TV content and TV-related content available on the Web. With the evolution of TVs and RSs, the diversity of recommender systems for TV has increased substantially. In this direction, it is worth mentioning that we consider “recommender systems for TV” as those that make recommendations of both TV-content and any content related to TV. Due to this diversity, more investigation is necessary because research on recommender systems for TV domain is still broader and less mature than in other research areas. Thus, this literature review (LR) seeks to classify, synthesize, and present studies according to different perspectives of RSs in the television domain. For that, we initially identified, from the scientific literature, 282 relevant papers published from 2003 to May, 2015. The papers were then categorized and discussed according to different research and development perspectives: recommended item types, approaches, algorithms, architectural models, output devices, user profiling and evaluation. The obtained results can be useful to reveal trends and opportunities for both researchers and practitioners in the area.