Developing Sequence-Aware Recommender Systems with Hidden Markov Models

Abstract

Recommender systems (RS) are one of the most successful applications of data mining and Machine learning technology in practice. It is generally based on the matrix completion problem formulation, where for each user-item pair only one interaction (rating) is considered. In many application domains, multiple user-item interactions of different types can be recorded over time. And a number of recent works have shown that this information can be used to build richer individual user models and to discover additional behavioral patterns that can be leveraged in the recommendation process in what is called Sequence-Aware Recommender Systems (SARS). This thesis aims to highlight this new kind of Recommender systems (SARS). In addition, a Hidden Markov Model (HMM) to develop SARS is studied. Results of the experiments on the Last.fm dataset, are presented using a different set of HMM parameters