Application of Improved DBSCAN Algorithm in the Plan Compilation Management
For further solving the problem of normative emergency plan, text mining should be combined with emergency plan compilation. Fixed "Threshold strategy is used by traditional DBSCAN, which would lead to the problem of cluster boundary wrong recognition. Improved DBSCAN algorithm is introduced in this paper. Least Square Fit is taken to fit plans similarity curve to find the best of initial "Threshold. According to the initial ", a new strategy is used to get dynamic "Threshold to improve the precision and recall. The simulations results show that the presented method is efficient for providing intelligent reference groups for government staff.