IPTV Data Reduction Strategy to Measure Real Users’ Behaviours
Nov 2012
We present a strategy that deals directly with the real IPTV data, which result from the interaction actions with the set- top boxes by customers. But these data have a very low level of granularity, complex and difficult to interpret. The approach is to transform the clicking actions to a more conceptual and representative level of the running activities. Furthermore, there is a significant reduction in the data cardinality, enhanced in terms of information quality. More than a transformation, this approach aims to be iterative, where at each level, we achieve a more accurate information, in order to characterize a particular behaviour. As experimental results, we present some application areas regarding the main offered features in this digital service: zapping behavior, DVR service usage.
