About

dbrec is a new kind of recommender system, entirely based on Semantic Web and Linked Data technologies, providing recommendations for 39,000 artists from DBpedia (and consequently from Wikipedia) based on the relationships that they hold together, either directly (e.g. having played together) or indirectly (e.g. playing the same genre, sharing a common record label or even being associated in a particular cultural movement).

Moreover, not only it provides recommendations, but it explains these recommendations by presenting the users how they have been computed. For instance, the following picture shows why the system recommends Big Brother and the Holding Company for Janis Joplin. For each part of the explanation, the system also presents the number of other artists and bands sharing that link in the whole dataset.

To use dbrec, simply use the search box on the left, wait for the autocompletion and validate your choice. If you know the english Wikipedia URL of an artist or band of your choice, e.g. http://en.wikipedia.org/wiki/Ramones, you can directly access it through its dbrec URL, e.g. http://dbrec.net/page/Ramones.. Once you access an artist page, the top 15 recommendations are listed and you can simply check the explanations for each of them by following the "Why are they related ?" link. In addition, you can share the recommendations on your favorite social networking services.

Technologies

dbrec makes a full-use of Semantic Web and Linked Data technologies in order to provide - and explain - its recommendations.

It uses DBpedia as a seed RDF dataset from which it provides recommendations for more than 30,000 artists thanks to a Linked Data Semantic Discance (LDSD) algorithm, computed using a set of SPARQL queries over the dataset. The results are modeled using the related LDSD ontology and made available to the users through the current user-interface using RDFa, so that new applications can be built on the top of it.

Data access

Recommendations provided by dbrec are freely-accessible (see license terms below) by being available as RDFa annotations in the pages using the LDSD ontology. While dbrec does not currently provide access to its full data via a SPARQL endpoint, this is something we envision for the feature, resources permitting.

Licence

dbrec content is available under the terms of the Creative Commons Attribution-Share Alike 3.0. All information about dbrec artists (pictures, labels and descriptions) come from DBpedia.

Credits

dbrec is designed and maintained by Alexandre Passant at DERI, NUI Galway. The project is supported by Science Foundation Ireland under grant number SFI/08/CE/I1380 (Líon 2).

We rely on the DBPedia data and 4store to compute the recommendations using the LDSD algorithm. The interface is build thanks to SPARQL, librdf, PHP and MySQL.

Contact

For any question about dbrec, please contact alexandre.passant[at]deri.org including [dbrec] in the subject line.