军事知识图谱的构建和检索方法研究

Research on the construction and retrieval method of military knowledge graph

针对军事信息系统中的信息过载现象,为了提高军事领域信息利用效率,提出了一种基于军事知识图谱的语义检索方法。首先,针对军事文本的特点,在基于转换器的双向编码器表示—条件随机场(bidirectional encoder representation from transformers-conditional random field, BERT-CRF)模型基础上添加了校对处理层,提升了军事实体的识别效率,构建了更全面的面向军事领域的知识图谱;然后,使用多相似度融合的语义计算方法对其进行语义分析;最后对检索结果进行相关性排序。实验结果表明该方法的查全率和查准率分别达到了85.68%和85.74%,能够满足军事领域下的信息检索要求。

Aiming at the information overload in military information system, a semantic retrieval method based on the military knowledge map was proposed to improve the efficiency of information utilization in the military field.Firstly, according to the characteristics of military texts, a proofreading processing layer was added to the BERT-CRF(bidirectional encoder representation from transformers-conditional random field) model to improve the recognition efficiency of military entities, and a more comprehensive knowledge map for the military field was constructed.Then, the semantic computing method of multi-similarity fusion was used to carry out the semantic analysis.Finally, the retrieval results were sorted by relevance.Experimental results show that the recall and precision of this method are 85.68% and 85.74% respectively, which can meet the requirements of information retrieval in the military field.