Author(s)

Jingya Huang

Corresponding Author

Jingya Huang

ABSTRACT

Reputation evaluation is the public's credit evaluation of a series of historical behaviors of individuals, organizations or enterprises. It is a comprehensive evaluation formed by obtaining relevant feedback information from the public in social networks. Its key is to comprehensively quantify the accumulation of historical behaviors of entities. Feature extraction and multi-objective machine learning algorithm is proposed based on multi-objective cooperative FTEA. The algorithm finds out its kernel attribute by feature extraction of multi-attribute of learning samples, and the kernel attribute and other non-kernel attributes form an attribute group, thus improving the classification accuracy. This paper presents a fuzzy reputation level evaluation algorithm (FTEA) for generating evaluation rules and reputation levels. The algorithm adopts the rule-based machine learning method, which has the ability of self-learning from a large number of input data to obtain evaluation rules. In reputation management, the model adopts the cluster head management mechanism, which alleviates the problem of slow reputation convergence caused by complex computing, and is suitable for the network expansion needs. In the process of reputation update, the reputation information update of related nodes drives the reputation update of each role node.

KEYWORDS

Feature extraction, Machine learning, Reputation, Dynamic reputation