Evolutionary Computing Approaches to Multi Criteria Recommender Systems
Loading...
Date
item.page.authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
One of the most important challenges facing us nowadays is to customize information based
newlineon user preferences. In order to accomplish this objective, recommender system (RS) came
newlineinto existence that provides personalized recommendations to users through different decisionmaking
newlineprocess based on user s likings and constraints. RS became an important research
newlinearea in last two decades and it has been applied successfully in many domains, such as videos
newline(YouTube), movies (Netflix), songs (Last.FM), books (Amazon), online social networks (Facebook,
newlineTwitter), news articles (Globo.com), hotels (Goibibo) etc.
newlineMost of the RSs consider overall ratings to generate recommendations to users. Considering
newlineonly overall rating is not sufficient to capture user preferences efficiently because users
newlinemay like or dislike any item based on various criteria. Therefore, recommender systems based
newlineon overall rating may not perhaps produce quality recommendations to users based on their
newlinepreferences. So, instead of considering only overall rating in RS, it would be better to incorporate
newlinecriteria ratings while generating recommendations. Hence, multi-criteria recommender
newlinesystems (MCRSs) came into existence to provide quality recommendations to users by considering
newlinevarious criteria ratings. However, rating data may be more sparse. Over the past decade,
newlineMCRS is classified into two categories namely similarity-based approach and model-based approach.
newlineIn similarity-based approaches, aggregation of these criteria ratings is an important
newlinechallenge for generating recommendations to users. Most of the research in MCRS is focused
newlineon estimating users preference but none has focused on item utility. However, item utility
newlinecan be a good factor in consideration of quality recommendations to users. In model-based
newlineapproaches, elicitation of preference function is a major concern, that can reflect a relationship
newlinebetween overall and criteria ratings.
newline