基于模态匹配的校园网信息实时过滤方法

互联网的信息良莠不齐，为避免校园网发送和接收不良的信息，通常都需要实时监测并对敏感关键词实行过滤。提出利用模态匹配的方法进行信息过滤，研究一种新的实时监测与过滤网络信息的方法。通过提取实时监测到数据信息的特征值并转化为向量表示，在MAC算法的基础上，改进其算法，把监测到的数据信息的特征向量与需要过滤的源信息库的特征向量进行模态匹配，以得到的匹配度作为判断实现信息过滤的依据。通过在网络实训室模拟广域网进行过滤实验，实验结果表明模态匹配的实时网络信息监测并过滤的方法可行，准确率能达到应用需求，模态匹配造成的延迟时间在可接受的范围内。

Methods About Real-time Monitoring and Filtering of Network Information Based on Modal Matching

Because good and bad information is mixed on network, it is necessary to monitor and filter sensitive key words in many cases. For the target, a new method based on modal matching is in process of studying. By means of extracting eigenvalue of real-time monitored information, vectorizing it, and then improving the algorithm based on Modal Assurance Criterion （MAC） algorithm, eigenvectors of information monitored are matched with those required to be filtered of source repository, in order to get a matching degree as a judgement of information filtration. In accordance with results of filtration experiments in an analog WAN of a network training room, the method of real-time monitoring and filtering of network information based on modal matching is feasible. The accuracy can be accepted for application, and delay time caused by modal matching is in acceptable range.