摘要
基于当下灭火救援领域中的突发事件,提出灭火救援专业知识智能匹配算法,该算法基于自然语言处理和注意力机制计算案件描述和消防预警信息之间的语义关系,从而实现相关灭火救援专业知识的匹配。首先基于自然语言处理的方法学习句粒度级别的语义信息,然后基于注意力机制学习词粒度级别的语义信息,最后基于两个级别语义信息的交互,根据信息之间局部差异推断两个句子之间的关系。试验结果表明:该算法具有优异的性能,能够同时从词和句两个粒度上更准确地理解句子,实现基于案件描述的灭火救援专业知识智能匹配。
Based on the current emergencies in the field of firefighting and rescue, an intelligent matching algorithm for firefighting and rescue expertise is proposed. The algorithm calculates the semantic relationship between case description and firefighting knowledge based on natural language processing and attention mechanism, realizing the matching of related firefighting and rescue knowledge. First learn semantic information at sentence granularity level based on natural language processing, then learns semantic information at word granularity level based on the attention mechanism, and finally, based on the interaction of two levels of semantic information, the relationship between sentences is inferred based on the local differences of the information. Experimental results show that the algorithm has excellent performance. At the same time, it can understand sentences more accurately from the two granularities of words and sentences, realize intelligent matching of firefighting and rescue professional knowledge based on case descriptions.
作者
张毅
Zhang Yi(Shanghai Baoshan Fire and Rescue Division,Shanghai 201999,China)
出处
《消防科学与技术》
CAS
北大核心
2022年第10期1472-1476,共5页
Fire Science and Technology
关键词
灭火救援
智能匹配
自然语言处理
注意力机制
语义信息
fire rescue
intelligent matching
natural language processing
attention mechanism
semantic information
