Research Of Bayesian Algorithm In The Application Of Intelligent Terminal Information Filering

With the rapidly development of Internet, intelligent terminal information system is accepted by the majority of Internet users for the advantages of easy to use, fast, cheap, reliable, and has become the most popular means of information exchange. Intelligent terminal system brings convenience, but can bring spam at the same time, which has brought great harm. Therefore, making the use of filtering technology, barring the intelligent terminal information and solving the problem of intelligent terminal is of great significance.Bayesian filtering technology is applied in text classification and so on. Intelligent terminal information filtering is essentially a text classification problem, and the Bayesian algorithm has a very good application in the field of intelligent terminal information filtering. As a starting point, the thesis analyzes the information content of the intelligent terminal filtering technology, summarizing and analyzing the status quo, and proposes a series of improvements based on Bayesian filtering method. On this basis, the thesis studies Bayesian algorithms in detail, and proposes three areas of improvement ideas. For text, thesis mainly choose to use the fingerprint as characteristics; use conditions as the characteristics of expressed; the dynamic adjustment of the threshold algorithm with the deepening of learning. Based on these improvements, the improve Naive Bayesian filter. Through analyzing the characteristics of the message structure, we propose integrated weighted model for different from ordinary text, to take full advantage of the structure of the message. Then, on this basis, the thesis uses the message header and message body based on the integration of the weighted model to make models. With integration between the two results, the finally intelligent terminal information by a weighting filter and learn two Bayesian extended model of minimum risk Bayesian and active learning Bayesian. Based on the comparison of experimental results, we get the optimal conditions of two extended model and improve mail filtering algorithm. The test results show that, compared with the classic Bayesian filter, the filtering effect of the Bayesian filter with a combination of the above improved and expanded technology the paper proposes, has a high accuracy. The research of this paper can provide a theoretical basis for the design and implementation of similar systems.