A Multi-Dimensional Feature Combination Method for Identifying Threat Intelligence Entity Relationships

At present, most of the massive threat intelligence analysis relies on manual processing and evaluation,which cannot achieve real-time update of huge data, resulting in differences in unstructured text data and difficulty in extracting entity and relationship information. In view of the above problems, this paper proposes a multi-dimensional feature combination threat intelligence entity relationship recognition method. Through entity data feature extraction and feature matching, the extracted atomic feature function and the predefined atomic feature combination function of the text data content of the article are set to realize the multi-dimensional atomic feature combination,Then,the multi-dimensional atomic feature combination vector and relationship vector are used to construct a feature combination relationship graph, which is combined with graph attention network to update relationship nodes.Finally, the entity relationship prediction layer achieves effective recognition of threat intelligence entity relationships for the content of article text data. The results of the comparison model indicate that the model has better ability to identify threat intelligence entity relationships.