A Real-Time Data Filter for Interactive-TV Rating System

Abstract

The set-top box (STB) connected to TVs are used to gather viewing data, which then is transmitted to a backend platform. It is very hard to analyze the viewing behavior since the viewing data are very huge and contain noise. This paper presents a real-time interactive-TV rating system for monitoring the trend of user viewing behavior. The system collects viewing behavior data from CATV system vendors, analyzes the data, and then provides real-time analysis diagrams. The authors also proposed two phases filtering algorithm that can efficiently check and filter the invalid viewing behavior according to the effective data filter rules. The algorithm can daily filter 27 million viewing behavior data and generate 7.6 million effective data. It can reduce the need of storage space and increase the analysis speed. In this works, we develop a real-time interactive-TV rating system for monitoring the trend of user viewing behavior and provide the real-time ratings for the last three minutes watched.