A Novel Semantic Cohesion Approach for Chinese Airworthiness Regulations: Theory and Application

Abstract

Airworthiness regulation documents are critical, which contain massive safe constraints for system design and airworthiness certification. The textual information of airworthiness regulations is organized and stored in a hierarchical and scattered form due to the unique provision structure. The linguistic form of legal provision brings the problems of incompleteness semantic information and syntactic structure to each legal clause, which limits the application of Natural Language Processing (NLP) technology in the field of airworthiness safety. In this article, a novel theory of semantic cohesion is proposed for Chinese airworthiness regulations, in which four critical elements are contained, including definition, model, theorem and rules. The definition of attributive directed graph model is proposed to achieve the graphical representation of airworthiness regulation texts. Moreover, the graphical issue of regulation texts is proved in accordance with the characteristic of hierarchical provision structure based on graph theory, and the rules of edge structure construction are provided. Based on the theory, a new node content link algorithm is proposed to achieve the tasks of semantic cohesion and structure conversion for Chinese airworthiness regulation texts. According to the hierarchical structure characteristic of regulation texts, the algorithm can convert the provision structure to the general narrative linguistic form by constructing edges with semantic cohesion relation. As a result, provision sentences with complete semantic information and syntactic structure can be generated. The algorithm has been deployed on the current 119 airworthiness regulation texts to verify the validity and feasibility. The experimental results show that the algorithm achieves efficient performance in the tasks of structure conversion and semantic cohesion, and the algorithm performs reliably with an accuracy rate of 100%. The problem of application limitation for NLP technology to Chinese airworthiness regulation texts has been resolved, which promotes the intelligent and automatic development of airworthiness safety.