Gabriel Sepúlveda, Denis Parra.
JI3 2017, number 9, pages 92-103.
ABSTRACT
Within recommendation tasks, either in the field of ratings prediction or in rankings development, there are two fundamental components allowing them to be carried out. One is a set of data that relate the level of affinity expressed by certain users towards certain elements or items. The other is a group of algorithms capable of converting the data into information that predicts degrees of affinity or rejection that users may experience towards items belonging in the subset of unknown elements. To bring these two components together, there are multiple software tools that manage to solve the recommendation problem without having to go into the intricacies inherent in programming algorithms, but do not have a complete battery of tools for conducting experiments and analyses in a single environment. The pyRecLab library was developed in order to overcome these shortcomings by providing easy-to-use recommendation methods, which are expected to expand their coverage over time.