Establishment of a Copyright Regulations Knowledge Base and Development of a Case Recommendation System Utilizing Transformer and Graph Convolutional Network

Abstract

To disseminate copyright regulations and address issues related to lapse, inexactitude, and inadequacy in the eminence of case data, we propose an initial methodology for the construction of textual content based on copyright regulations and cases. This methodology involves the processing of regulatory and case information, followed by the exploration of interrelationships. Subsequently, we use the Transformer algorithm for semantic information processing to extract nuances like conceptual terminology, pivotal keywords, and elucidating annotations from cases. This effort facilitates the creation of a concept index for cases, promoting case archiving. Concurrently, we introduce a methodology relying on keywords for the extraction of legal or case-related concepts. Recognising the multifaceted nature of cases with diverse sub-nodes, we propose a feature alignment approach grounded in Graph Convolutional Networks (GCN). This innovation serves as the basis for logically acclaiming copyright regulations within our knowledge framework. Empirical validations accentuate the effectiveness of our case recommendation system, showcasing an accuracy rate of 86.5%. Additionally, our compilation of copyright regulatory knowledge garners outstanding accolades in subjective evaluations.