A Novel Filtering Recommendation Algorithm for User Emergency Information Adoption

Abstract

Emergency case data resources are widely distributed and heterogeneous. At the same time, the command of emergency field needs the cooperation of multiple departments. Therefore, it is urgent to establish an emergency analysis and mining platform, realize the sharing and collaboration of emergency data resources among multiple departments, and assist emergency command and scheduling. According to the actual situation of the current emergency, a similarity measure method (TCRD) is proposed to solve this problem by adding temporal information to reflect information adoption, which integrates user context information and temporal information. Firstly, the temporal information of historical adoption behavior is expressed as a binary coded characteristic matrix, and then the characteristic matrix is mapped into a feature vector by using restricted Boltzmann machine, and finally added to the similarity measurement formula. The improved TCRD method can measure the similarity more accurately, and further improve the quality of emergency information adoption recommendation results.