SPrank

• DOI: 10.1145/2899005
• Publication Date: 2016-09-21T12:42:46Z
• Authors: Tommaso Di Noia; Vito Claudio Ostuni; Paolo Tomeo; Eugenio Di Sciascio
• Keywords: Linked Data; Learning to Rank; Rank (graph theory)

In most real-world scenarios, the ultimate goal of recommender system applications is to suggest a short ranked list items, namely top- N recommendations, that will appeal end user. Often, problem computing recommendations mainly tackled with two-step approach. The focuses first on predicting unknown ratings, which are eventually used generate recommendation list. Actually, task can be directly seen as ranking where main not accurately predict ratings but find best-ranked items recommend. this article we present SPrank, novel hybrid algorithm able compute exploiting freely available knowledge in Web Data. particular, employ DBpedia, well-known encyclopedic base Linked Open Data cloud, extract semantic path-based features and learning-to-rank fashion. Experiments three datasets related different domains (books, music, movies) prove effectiveness our approach compared state-of-the-art algorithms.