Cross-system heterogeneous data fusion characterization method

The invention discloses a cross-system heterogeneous data fusion characterization method, and relates to the field of electric power information systems. An electric power system has many heterogeneous data systems and devices, heterogeneous data between different systems is difficult to fuse, and the working difficulty is increased for specific data analysis and anomaly detection tasks. According to the method, the data association graph between the two heterogeneous data systems is constructed, then the meta-path is automatically generated based on the data association graph, and finally, under the constraint of the meta-path, a node sequence is obtained based on random walk to maximize the conditional probability of occurrence of each node, so that the final embedding result of each node can be obtained. Therefore, fusion characterization of the data is realized. And cross-system heterogeneous data fusion representation is effectively realized, and the application difficulty is reduced. 本发明公开了一种跨系统异构数据融合表征