关键词: 消防应急救援规范, 自动结构化, 知识图谱(KG), 语言树状结构解析(LTP)模型, 语义依存分析(SDP)

Abstract:

To enhance processing efficiency and semantic mining depth for fire emergency response regulatory provisions, an automatic structuring method integrating LTP model and KG technology was proposed in this paper. First, the SDP on regulatory sentences was performed by the LTP model to uncover in-depth semantic associations. Second, a keyword extraction strategy was further developed. Standard terminology was identified through structural parsing, while cross-regulation high-frequency core terms were selected via document frequency statistics to effectively capture critical information. Then, based on semantic dependency relationships and core argument features, regulatory provisions were classified into four categories: citation, requirement, parameter and evaluation, and the automatic extraction rule base was established. Finally, this method was applied to 81 fire protection regulatory provisions. A total of 330 terms and 2 170 repeated words were extracted, resulting in a KG comprising 15 940 nodes and 286 294 relationships. The results show that precise querying using the Neo4j graph database reveals that parameter-based provisions account for the highest proportion (80.46%), indicating their foundational role in the regulatory system. This verify this method is effective for the classification and semantic mining of regulatory provisions.

Key words: firefighting emergency rescue regulations, automatic structuring, knowledge graph (KG), linguistic treelet parsing (LTP) model, semantic dependency parsing (SDP)