Research on Improved CLIQUE Partial Order Algorithm Weight

Aiming at the problem that the existing CLIQUE clustering algorithm does not consider the feature weight, which leads to the low accuracy, a weighted improvement method combined with POset idea is proposed. First, obtain the weight order of features. The original data are then weighted in partial order. Finally, the traditional CLIQUE algorithm is used to cluster the weighted data. This method can effectively integrate the weight information into the algorithm when only the feature weight order is obtained. The experimental results show that the clustering accuracy has been significantly improved, which fully reflects the role of feature weight. At the same time, the idea of POset can effectively integrate expert information. The representation of nearest neighbor elements in Hasse graph can enhance the robustness of clustering results. This is an effective method to improve CLIQUE clustering algorithm.