Extracting Domain Knowledge Elements of Construction Safety Management: Rule-Based Approach Using Chinese Natural Language Processing

Abstract

The literature and practices of construction safety management have highlighted the importance of domain knowledge. Effectively extracting the domain knowledge elements (DKEs) of construction safety management remains a challenging task. To address this problem, this paper develops a rule-based natural language processing (NLP) approach for extracting DKEs from Chinese text documents in the domain of construction safety management. First, a linguistic pattern of DKEs was constructed according to lexical analysis and syntactic dependency parsing. Then, the extraction rules and workflow paths were established and tested. The results indicated that most DKEs in the domain of construction safety management are composed of specific compound parts of speech (nouns and noun phrases), specific word dependencies (attribution, verb-object, subject-verb, preposition-object, and coordinate relationship), and words of specific lengths (two to six Chinese characters). This work is the first to reveal the Chinese linguistic patterns and linguistic features of DKEs in the domain of construction safety management. The findings of this study can facilitate the establishment and supplementation of domain lexicons and knowledge-based safety management systems and can guide safety training for construction safety management.