Athirai Aravazhi Irissappane

Filtering unfair ratings from dishonest advisors in multi-criteria e-markets: a biclustering-based approach

In multiagent e-markets, trust between interaction partners (buying agents and selling agents) is vital for any transaction to be successful. Given the difficulty for a buyer to directly judge the quality (trustworthiness) of a seller for a transaction, a buyer also seeks opinions from other buyers (called advisors) in the marketplace to determine the seller’s trustworthiness. However, advisors may act dishonestly by conveying misleading information about the seller. We propose a novel approach to identify such dishonest advisors, while evaluating a seller’s trustworthiness on multiple criteria. It is based on a biclustering method which clusters honest advisors on different criteria. Correlation between advisors’ ratings to various criteria is used as additional information to accurately filter dishonest advisors. A transitive mechanism is also employed in the biclustering process to cope with rating sparsity. Further, we introduce a parallelization technique to reduce the time complexity involved in the biclustering process. Detailed experiments in simulated environments demonstrate the robustness of the proposed approach against strategic attacks from dishonest advisors. Evaluation on three real datasets confirms the effectiveness of our approach in real environments.

A case-based reasoning framework to choose trust models for different E-marketplace environments

The performance of trust models highly depend on the characteristics of the environments where they are applied. Thus, it becomes challenging to choose a suitable trust model for a given e-marketplace environment, especially when ground truth about the agent (buyer and seller) behavior is unknown (called unknown environment). We propose a case-based reasoning framework to choose suitable trust models for unknown environments, based on the intuition that if a trust model performs well in one environment, it will do so in another similar environment. Firstly, we build a case base with a number of simulated environments (with known ground truth) along with the trust models most suitable for each of them. Given an unknown environment, case-based retrieval algorithms retrieve the most similar case(s), and the trust model of the most similar case(s) is chosen as the most suitable model for the unknown environment. Evaluation results confirm the effectiveness of our framework in choosing suitable trust models for different e-marketplace environments.

A Context-Aware Framework for Detecting Unfair Ratings in an Unknown Real Environment

Reputation systems are highly prone to unfair rating attacks. Though many approaches for detecting unfair ratings have been proposed so far, their performance is often affected by the environment where they are applied. For a given unknown real environment, it is difficult to choose the most suitable approach for detecting unfair ratings as the ground truth data necessary to evaluate the accuracy of the detection approaches remains unknown. In this paper, we propose a novel Context-AwaRE (CARE) framework, to choose the most suitable unfair rating detection approach for a given unknown real environment. The framework first identifies simulated environments, closely similar to that of the unknown environment. The detection approaches performing well in the most similar simulated environments are then chosen as the suitable ones for the unknown real environment. Detailed experiments illustrate that the CARE framework can choose the most suitable detection approach to accurately distinguish fair and unfair ratings for any given unknown environment.