An Agent Based Middleware for Privacy Aware Recommender Systems in IPTV Networks

Abstract

IPTV providers keen to use recommender systems as a serious business tool to gain competitive advantage over competing providers and attract more customers. As indicated in (Elmisery, Botvich 2011b) IPTV recommender systems can utilize data mashup to merge datasets from different movie recommendation sites like Netflix or IMDb to leverage its recommender performance and predication accuracy. Data mashup is a web technology that combines information from multiple sources into a single web application. Mashup applications created a new horizon for different services like real estate services, financial services and recommender systems. On the other hand, mashup applications bring about additional requirement related to the privacy of data used in the mashup process. Moreover, privacy and accuracy are two contradicting goals that need to be adjusted for the spread of these services. In this work, we present our efforts to build an agent based middleware for private data mashup (AMPM) that serve centralized IPTV recommender system (CIRS). AMPM is equipped with two obfuscation mechanisms to preserve privacy of the dataset collected from each provider involved in the mashup application. We present a model to measure privacy breaches. Also, we provide a data mashup scenario in IPTV recommender system and experimentation results.