Robust Truth Discovery Scheme Based on Mean Shift Clustering Algorithm

Data conflict is an inevitable problem in data collection due to the limitations in the real world. How to determine the most reliable data among many collected data is a problem worth studying. Truth discovery is a widely used resolution to integrate heterogeneous data. Existing truth discovery schemes mainly aggregate data from all sources to obtain the truth. However, it is intuitive that exceptional data should be excluded before aggregation, and how to identify exceptional data is also difficult. To tackle this problem, we use the mean shift clustering algorithm to remove exceptional data and obtain the truth value. Experiment results indicate our scheme performs well in various conditions.