INVESTIGATING THE NUMBER OF USERS AND MONTHS TO MAKE TULUNGAN EFFECTIVE AGAINST SELF-PROMOTING USERS

Tulungan is a reputation system for collaborative web filtering that is both consensus-independent and self-promoting resistant. Simulation results show that Tulungan requires less than 50% good users in order to become effective against malicious and self-promoting users (i.e., bad users). Being effective means that it is able to give high reputation values to good users relative to bad users. In addition, the effectiveness of Tulungan is also confirmed by comparing the number of correct URL categorizations that it made against the wrong ones. This is in contrast with other reputation systems that require at least 50% good users in order to be effective against malicious users and even a greater percentage of good users to work against self-promoting users. Although previous studies show the consensus-independent and self-promoting resistant properties of Tulungan, the simulation covers only a fixed number of total users and months. This paper presents additional simulation involving Tulungan that confirms the number of users and months needed for it to become effective. Results show that the reputation values of good users suffer if it involves less than 200 users and 2 simulated months.