Recommender systems in libraries

iniciatives and proposal of a theoretical hybrid model

Abstract

Recommender systems are software tools that have as a goal the generation of an estimate of usefulness of items for a certain user, as well as analyzing interactions between users and items. They are divided between classic recommender systems, which use only one type of filtering technique, and hybrid, which use two or more types of filtering techniques allied to a hybrid recommendation technique. The goal of this paper is to introduce the recommender systems used in libraries and propose a theoretical hybrid recommender system model, adapted to the context and reality of libraries. For this purpose, a qualitative, exploratory, descriptive and bibliographic approach was used to explore the theme of recommender system and their use in libraries as an answer to users’ growing informational demand. A conclusion was made that most of the recommender system initiatives that are in use in libraries today are based on the classic recommender model and use only two types of filtering techniques – by content and collaborative. Therefore it is suggested that future research needs to try to comprehend the reason why the classic recommender model is used more, in detriment to hybrid recommender system, and that a bigger focus on the development and application of hybrid recommender system for libraries is needed, given its advantages when compared to the classic model. It is also suggested that further studies on other types of filtering techniques be made, given their potential to generate new recommendation strategies in libraries.