Combining Multiple Forms of Evidence While Filtering

Yi Zhang, Jamie Callan

Abstract: This paper studies how to go beyond relevance and enable a filtering system to learn more interesting and detailed data driven user models from multiple forms of evidence. We carry out a user study using a real time web based personal news filtering system, and collect extensive multiple forms of evidence, including explicit and implicit user feedback. We explore the graphical modeling approach to combine these forms of evidence. To test whether the approach can help us understand the domain better, we use graph structure learning algorithm to derive the causal relationships between different forms of evidence. To test whether the approach can help the system improve the performance, we use the graphical inference algorithms to predict whether a user likes a document based on multiple forms of evidence. The results show that combining multiple forms of evidence using graphical models can help us better understand the filtering problem, improve filtering system performance, and handle various data missing situations naturally.