Research On Technologies Of Malicious Behavior Detection In Expert Evaluation System

Project evaluation system is to determine the subject matters prescribed goal of suitability, adequacy and effectiveness of the activities and the design of system, under the background of mobile Internet, the traditional mode of project evaluation experts have been unable to adapt to the project evaluation rating system efficiency, the demand of the reusable, and project evaluation system based on network can meet the above requirements. But at the same time due to the scoring process is subjective, is unable to avoid malicious behavior, namely malicious score. Of the existence of malicious rating evaluation system of justice, fairness and normative cannot be guaranteed, in order to ensure the integrity evaluation system, the standard evaluation system, make evaluation project treated’, you must identify malicious score.The purpose of this paper is to solve these problems, in the process of recognition score malicious rating, its essence is through the collaborative filtering algorithm implementation recommendation system, through the score before in the history of the clustering, experts by other experts in the relevant classes of score data to evaluate the whether the expert scoring for malicious.In this paper, the main research content mainly includes the following aspects:A recommendation system of several algorithms are analyzed, system, collaborative filtering recommendation, based on the utility is recommended, based on knowledge recommendation, rule-based recommendation, based on the content recommended, finally through comprehensive comparison is suitable for expert evaluation system to identify malicious scoring algorithm is:collaborative filtering algorithm.Second, compared the user-based collaborative filtering algorithm and the collaborative filtering algorithm based on item, in-depth analysis of the core and key technologies of collaborative filtering algorithm, finally through comprehensive comparison is suitable for expert evaluation system to identify malicious scoring algorithm is:the collaborative filtering algorithm based on the user. Three, implementation is suitable for expert evaluation system to identify malicious score of collaborative filtering algorithm based on the user, in order to realize the collaborative filtering algorithm based on user identification of malicious score, first to calculate similarity between experts, as the basis of expert group clustering, secondly establish evaluation matrix, again through the target experts interested in any item of degree prediction calculated recommended prediction score, at last, by MAE to verify the accuracy of the collaborative filtering algorithm.At the end of the paper on some unit of science and technology project evaluation system, realize the collaborative filtering algorithm to identify malicious score of evaluation system, with some experts as a test of users, the actual evaluation scores and prediction score evaluation index is relatively close to illustrate the collaborative filtering algorithm based on the user in the prediction of malicious rating performance outstanding, malicious rating by the recognition of the implementation of evaluation system is proved the validity of the practice, this paper finally through comparing the difference rate, identify the malicious rating, error detection rate is low, good recognition result.