基于词库过滤与分布式词向量的安全生产风险问题智能匹配算法分析研究

Analysis and Research on Intelligent Matching Algorithm for Safety Production Risk Problem Based on Thesaurus Filtering and Distributed Word Vector

针对现有安全生产风险问题匹配技术无法考虑词义或避免描述噪声影响的问题,提出一种基于词库过滤与分布式词向量的智能匹配算法。该方法通过词频统计和专家判断相结合的方式,建立特征词库与过滤词库,并基于词库对安全生产风险问题描述进行特征提取以及噪声过滤,结合分布式词向量进行基于词义的相似度计算,实现安全生产风险问题的精确匹配。研究证明,该方法能够有效提升安全生产风险问题相似问题匹配精度,在处理新发现的安全生产风险问题时为安监人员快速匹配相似问题提供参考,为企业安全生产风险问题的处理管控提供指引。

In response to the problem that existing safety production risk matching technology cannot consider word meaning or avoid the impact of descriptive noise,an intelligent matching algorithm based on thesaurus filtering and distributed word vectors is proposed.This method combines word frequency statistics and expert judgment to establish a feature thesaurus and a filtering thesaurus.Based on the thesaurus,feature extraction and noise filtering are performed on the description of safety production risk problems.Calculates similarity is taking based on word meaning by combining distributed word vectors,to achieve accurate matching of safety production risk problems.Experimental results have shown that this method can effectively improve the matching accuracy of similar problems in safety production risk problems.It provides reference for safety supervision personnel to quickly match similar problems when dealing with newly discovered safety production risk problems,and provides guidance for the management and control of enterprise safety production risk problems.