Research on internet information mining based on agent algorithm

With the rapid development of information technology, especially network technology, people's ability to collect, store and transmit data are increasing. The data have exploded in an explosive manner. In sharp contrast, the ability to make valuable data for decision making is very poor. In this paper, data mining is the most basic problem. In order to overcome the shortcomings of the traditional clustering algorithm for k-means clustering, it is difficult to determine the initial clustering center and the k-means algorithm is improved. When determining the initial K-, the convergence factor is improved and the global optimum is achieved, so as to realize the determination of clustering center. By using improved k-means algorithm to approximate the criminal data, the validity of this method is verified.