Combining Multiple Evidence from Different Relevant Feedback Networks