Recommender systems are applied in a multitude of spheres and have a significant role in reduction of information overload on those websites that have the features of voting. Therefore, it is an urgent need for them to adapt and respond to immediate changes in user preference. To overcome the shortcomings of each individual approach to design the recommender systems, a myriad of ways to coalesce different recommender systems are proposed by researchers. In this chapter, the authors have presented an insight into the design of recommender systems developed, namely content-based and collaborative recommendations, their evaluation, their lacunae, and some hybrid models to enhance the quality of prediction.
TopIntroduction
Over the past few decades, Web use has made its impact on various arenas like e-commerce, social networking sites, and digital libraries. However, the irony is that an enormous amount of available information is not accessible to the users since they are unaware that it exists. It is therefore necessary to broadcast in an efficient manner the relevant information to the Web users. This situation offers a very attractive framework for researching in the form of new accurate and efficient techniques designed to access this information (Campos, Luna, & Huete, 2008). Recommender Systems (RSs) are a type of information filtering system that gives advice on products, information, or services that a user may be interested in. They assist users with the decision making process when choosing items with multiple alternatives. RSs are based on human social behavior where opinions and tastes of known acquaintances are taken into consideration while making decisions. RSs generate personalized recommendations incorporating various mathematical, probabilistic and soft computing techniques. Movie recommendation websites are probably the most well-known cases to users and are without a doubt the most well studied by researchers (Konstan, Miller, & Riedl, 2004; Antonopoulus & Salter, 2006; Yamada & Li, 2004) although there are many other fields in which RS have great and increasing importance, such as e-commerce (Jinghua, Kangning, & Shaohong, 2007), e-learning (Denis, 2007; Bobadilla, Serradilla, & Hernando, 2009), and digital libraries (Porcel & Herrera, 2010; Porcel, Moreno, & Herrera, 2009). Added to the meteoritic rise in dissemination of information, this decade has seen the development of many software products such as Web 2.0 that add to the generation of recommendations very abruptly. Therefore, it is a challenge for the researchers to design RS that can take into account many demographic, contextual and preference factors and predict useful items more efficiently. The core of a RS is its filtering algorithms namely knowledge, demographic, content, utility, collaborative-based or their combination, which when applied are called hybrid RS. Demographic based algorithm is established on the assumption that individuals with certain common personal attributes (sex, age, country, etc.) will also have common preferences, whilst content-based filtering recommends items similar to the ones the user preferred in the past. Utility-based and knowledge-based recommenders do not attempt to build long-term generalizations about their users, but rather base their advice on an evaluation of the match between a user’s need and the set of options available. Utility-based recommenders make suggestions based on the computation of the utility of each object for the user. However, the issue is how to create a utility function for each user. Currently, Collaborative Filtering (CF) is the most commonly used and studied technique based on the principle in which in order to make a recommendation to a given user, it first searches for the users of the system who have voted in the most similar way to this user, to later make the recommendations by taking the items most highly valued by the majority of their similar users (Fernando, Hernando, & Alca, 2011).Various approaches to recommend the items in a domain are: