A Network Alert Degree Evaluation Model Based on Cooperative Computing

Oriented to the deficiency of cooperative analysis ability between different network domains in traditional topic detection and tracking (TDT), this paper proposes a network alert degree evaluation model based on cooperative computing (named CCAE). The model not only considers the influence of inner network domain, but also evaluates the cooperation impacts of inter-network domains. With the help of alert case knowledge base (ACKB), CCAE mines the alert information of each network domain by matching the alert cases with Web text. Through scheduling the local alert cases with high matching frequency between different network domains, CCAE effectively improves the ability of alert spreading analysis, optimizes the local alert case knowledge base and sharply shortens the time-consuming of alert detection. Considering the influence of page hyperlinks on alert spreading, CCAE retrieves the alert degree of the linked pages, and further realizes the evaluation in cooperative computing environment. For validating the performance of CCAE, we present a comparison experiment on a cosine classifier with TF-IDF in a corpus about "Campus Network Culture". The experimental results further validate the analysis ability of alert spreading and higher detection accuracy of CCAE.