Purpose – Web site recommendation systems help to get high quality information. The modelling of recommendation systems involves the combination of many features: metrics of quality, quality criteria, recommendation criteria, user profile, and specific domain concepts, among others. At the moment of the specification of a recommendation system it must be guaranteed a right interrelation of all of these features. The purpose of this paper is to model a web site quality‐based recommendation system by an ontology network. Design/methodology/approach – In this paper, the authors propose an ontology network based process for web site recommendation modelling. The ontology network conceptualizes the different domains (web site domain, quality assurance domain, user context domain, recommendation criteria domain, specific domain) in a set of interrelated ontologies. Particularly, this approach is illustrated for the health domain. Findings – Basically, this work introduces the semantic relationships that were used to construct this ontology network. Moreover, it shows the usefulness of this ontology network for the detection of possible inconsistencies when specifying recommendation criteria. Originality/value – Recommendation systems based on ontologies that model the user profile and the domain of resources to be recommended are quite common. However, it is uncommon to find models that explicitly represent the criteria used by the recommender systems, that express the quality dimensions of resources and on which criteria are applied, and consider the user context at the moment of the query.
International Journal of Web Information Systems – Emerald Publishing
Published: Nov 22, 2011
Keywords: Web sites; Information searches; Web information systems; Recommendations; Quality; Modelling; Ontology; Network