Building Recommendation Service with Social Networks and Semantic Databases

Building Recommendation Service with Social Networks and Semantic Databases

Sašo Karakatič (Institute of Informatics (FERI), University of Maribor, Slovenia), Vili Podgorelec (Institute of Informatics (FERI), University of Maribor, Slovenia) and Marjan Heričko (Institute of Informatics (FERI), University of Maribor, Slovenia)
DOI: 10.4018/978-1-4666-4490-8.ch008
OnDemand PDF Download:
No Current Special Offers


In this chapter, it is shown how useful user services can be created through the integration of social networks and semantic databases. The authors developed a recommendation service in a form of a Web-based application, where a user's interests are imported from social network Facebook and linked with additional data from open semantic database Freebase. Based on a custom implementation of k-nearest neighbors algorithm, the developed method is able to find recommendations based on users’ interests enriched with semantic information. The resulting list of found recommendations is then shown to the user in some basic categories like movies, music, games, books, and others.
Chapter Preview

Recommendation Service

The idea of predicting and recommending things to users based on their personal interest is not new. Similar commercial services already exist but are primarily aimed at recommending movies or music. One of them is Jinni1, which searches for movie recommendations, based on user’s profile created from semantic tags of liked movies. Another one is Pandora Radio2 online music service and works on a similar concept. User can respond with positive or negative feedback for each song that is played and Pandora then compiles playlist based on those feedbacks. Worth of mention is also service Last.fm3, where music recommendations are found through comparing user’s likes, similar basic step as is in our prototype service.

Complete Chapter List

Search this Book: