基于ST-DBSCAN算法的网络信息分类集成OA

In order to discover valuable information patterns from the vast and constantly changing network data,a network in-formation classification integration method based on spatial-temporal density-based spatial cluster of application with noise(ST-DBSCAN)algorithm is proposed.The relevant concepts of ST-DBSCAN algorithm are redefined to accurately capture the in-herent structure and distribution characteristics of data in the spatiotemporal dimension.Cluster processing is achieved by trav-ersing information points,checking spatiotemporal core points,constructing new clusters and identifying boundary points,and parameters are determined by drawing a spatiotemporal distance frequency histogram.After clustering,multiple spatially and temporally closely related clusters are obtained.To optimize the classification performance,ST-DBSCAN is combined with XG-Boost.By leveraging the supervised learning capability of XGBoost,a network information classification integration model is constructed through the Stacking integration learning strategy to achieve precise prediction of unknown information categories.The second-order Taylor expansion is used to optimize the loss function,improving the accuracy and efficiency of network in-formation classification intergation.The experimental results show that the proposed method has high clustering quality,clear information differentiation between clusters,noise distribution outside clusters,and the F1 value reachs about 0.96.The net-work information classification integration effect is good.