Preliminary Research On Contrastive Analysis Between Distributed Association Rules Algorithm And Distributed Decision Tree Algorithm

With the development of network technology and the wide adopted distributed computing environment, network security has been becoming people's main concern. Distributed Denial-of-Service(DDOS) attack is one of the dangerous network attacks, against which Distributed Intrusion Detection(DID) is thought to be one of the effective infrastructure protection methods. For data have been increasing exponentially, traditional protection measurements are not in a position to cope with all the network security problems under the distributed environment on a large scale. Combined with intrusion detection technology, datamining can make more efficient and precise analyses and judgements of the intrusive data.The thesis makes a brief introduction of the two algorithms: distributed association rules algorithm(FDM) and distributed decision tree algorithm(SPRINT) and a detailed comparision between the two, demonstrating their advantages and disadvantagefs respectively. Based on the two algorithms, the thesis makes some modifications and puts forward the improved FDM and SPRINT, developed by Microsoft Visual++6.0 and formatted by XML. By comparing the applicability and efficiency of the algorithms and the results of experiments therefrom, the thesis draws a conclusion that the improved FDM and SPRINT have a more efficient performance than their unimproved ones respectively; the improved FDM is more applicable than the improved SPRINT in datemining of distributed large dataset.