For the first set of experiments, we assume that a certain percentage of the recommenders are dishonest and launch bad mouthing attack by giving recommendations between 0.1 to 0.3.
To analyze the effectiveness of the proposed approach, we have simulated three inherent attack scenarios for recommendation models (bad mouthing, ballot stuffing, and random opinion attack).
The reason is that with higher trust deviation threshold, false recommendations from bad-mouthing nodes are not filtered out during the trust computation of evaluated nodes, which provides more opportunities to misbehaving nodes to remain undetected.
Using the mouth (gum, candy, mouthing non-food objects).
To analyze the effectiveness of the proposed approach, three inherent attack scenarios (bad mouthing, ballot stuffing, and random opinion attack) for recommendation models have been implemented in the above defined simulation environment.
In bad-mouthing attack, a misbehaving node propagates dishonest and unfair recommendations against an innocent node with a negative intention to confuse the trust model.
In bad mouthing attack, one or more entities falsely provide dishonest recommendation either to elevate trust values of malicious entities or to lessen the trust values of honest entities.
For example, the center's report was in part the basis of a Consumer Product Safety Commission recommendation that certain phthalates be removed from mouthing toys (such as teethers and rattles) due to concerns about children ingesting these chemicals.
It is also shown that for different attacks (bad mouthing, ballot stuffing, and random opinion attack), the proposed method successfully filters out dishonest recommendations.
The reason is that with higher deviation threshold, such as 0.5 and 0.6, false recommendation from bad-mouthing nodes having deviation of 60%% are only filtered out which causes legitimate nodes as misbehaving nodes, hence more false positives rate.
