Target threat level assessment is the key link of target situation cognitive reasoning. With the continuous development of artificial intelligence and big data technology, the demand for target threat level assessment in modern command and decision field is increasing. Most of the traditional target threat level assessment methods are based on expert knowledge, which have the problems of poor model robustness and low reasoning accuracy. In this paper, A method of target threat level assessment driven by knowledge and data is proposed. Target situation knowledge graph is constructed in combination with the task scenario, the multi-dimensional features of the target are represented as a unified vector, and the target entity and inter-entity relationship are embedded into the vector space by using the target situation knowledge graph, so as to realize the integration of different types of features into the unified space. Through the knowledge and data driven way to achieve the target threat level information mining. The effectiveness of the proposed method is verified by simulation experiments.