Construction Of Semantic Network Of Prototypical Events In Criminal Proceedings

Under the background of the data-driven and AI-driven era,the construction of smart justice is undergoing rapid development.The bottom-up constructing path through neural network and machine learning being somehow incapable of entirely describing the knowledge system of the applicational field in detail,and the semantic resources being scarce and low-structured,add up to a variety of problems to be remedied in this critical field.Grounded on this,the 3^rd generation of artificial intelligence,featuring the mixed driving force of both knowledge and data,combines these two paths in order to smoothly adapt to more applicational environments,in which a structured knowledge base mined and constructed from natural language is sure to be highly valued.The construction of a linguistic knowledge base begins with determining parameters including the oriented domain,structural features,functions and application values.Based on this,the domain-specific knowledge can then be mined and represented by means of specific methods of knowledge representation,which undoubtedly serve as the basis for knowledge acquisition and relevant integrated application.This study selects the criminal procedure law as the oriented domain,which,as a subordinate of jurisprudence,acts as a procedural law in the domestic legal system.It is a legal document on judicial procedures that aims to realize substantive rights and obligations,with apparently a handful of judicial organs,parties,temporal conditions,logical relations,etc.,involved.Different semantic relations between entities may directly affect the normative implementation of judicial procedure.In light of this,as a powerful method of knowledge representation,the semantic network is selected which takes semantic relations between nodes as the depiction object within the system and follows the trend of semantic system research.Having been introduced into natural language processing research as a relatively new linguistic theory,event semantics provides new assumptions for semantic analysis of knowledge units at all levels.Based on this theory,this study attempts to introduce the concept of the event into the constructing mechanisms of semantic network,which requires customization of node objects and semantic relational arcs inside the network.In depth from a meso perspective,given the consideration of thematic role theory and semantic relations,this study elaborates on the internal structural elements of events and the semantic relations between atomic events,resulting in the established two sets of meta-linguistic systems for the criminal judicial domain.Based on the systems,an event-based semantic analysis of the linguistic texts in criminal procedure law is carried out,and a corresponding semantic network is constructed.From the perspective of knowledge representation,the study attempts to optimize the representation mechanism of the network,broadens the depth and breadth of semantic processing ability,and hence can be regarded as a promising attempt at the construction of semantic resources.In summary,based on event semantics and semantic network,the present study attempts to build up event-oriented semantic resources in criminal procedure judicial practice.It can provide high-quality ensurance for the development of smart justice in China,bettering the current situation of insufficient construction of legal semantic resources and offering insights into the theory and methods of constructing large-scale semantic knowledge base in smart jurisprudence.