摘要
In the anti-spam area, distributed spam processing technology that based on reputation mechanism is a currently main focus in research. The raw reputation plays an important role in computing the finial reputation of the target node. In this paper, we analyze the ratio-based raw reputation generation approach in detail, point out its problems, and propose an aggregation-based raw reputation generation approach. Furthermore, we discuss how the number of evaluations and the method using to choose these evaluations impact on the result of raw reputation. Theoretical analysis and simulation results demonstrate that aggregation-based approach is much more effective and accurate than ratio-based approach. As long as we choose the proper number of evaluations, the results can accurately reflect behavior of the target node.
- 出版日期2009
- 单位南开大学