Evaluation of the information-spread model based on users' relevance

摘要

There are some methods for filtering the distributed information in the Internet, but no method based on the authority of information or transmitter. We consider that the significance of information is determined by user's reliability toward the author or the transmitter. Therefore, we proposed the information-spread model in which user re-transmits information to his neighbors or discard based on the relations to the neighbor. Each user does such process one by one. In this model, since the neighbors judge whether the information has a worth to read, the reliability of information can be estimated by the confidence to them. Those reliabilities can be used as filtering. We named this "Human Network based Filtering." The result of the experiment in this article shows that the relevance between users takes filtering effects.