Using trust-based recommender systems for personalized health content

Abstract

Recommender Systems intend to discover personalized items for a user based on item's descriptions and user's preferences. In Health era, Recommender Systems can be used to evaluate and extract useful personalized health content through mass of unreliable information on the web. Unlike many other types of recommendations, health recommendation extremely depends on emotional, physical and psychological matters of the user. Therefore using Personal Health Records or health profiles of user for personalized recommendation is inevitable but involves considerations about trust. In this work, we try to have a review of recommendation solutions for health domain. Furthermore, we analyze them considering their trust evaluation models, and propose an initial health recommender framework. This paper describes concepts of recommendation, medicine 2.0, and trust, reviews some solutions in these areas and then analyzes three most important attributes of web 2.0 (including trust and reputation, ontology and semantics, and social network collaboration),finally, proposes a framework for harvesting personal and trustable health content through social media to recommend. As the result, some feature of web 2.0 and their role in reviewed solutions are analyzed. It displays some weaknesses in using ontologies and semantic analysis. Moreover using different methods of trust management displays the different views of researchers about the notion of trust. Developing appropriate standards and ontologies, using semantic profile and content analysis to derive trust scores, and using limited personal health information solve current drawbacks partially.