Intelligent identification of trademark case precedents using semantic ontology

Abstract

A registered trademark distinctively identifies a company, its products or services. A trademark (TM) is a type of intellectual property (IP) which is protected by the laws in the country where the trademark is officially registered. TM owners may take legal action when their IP rights are infringed upon. TM legal cases have grown in pace with the increasing number of TMs registered globally. In this paper, an intelligent recommender system automatically identifies similar TM case precedents for any given target case to support IP legal research. This study constructs the semantic network representing the TM legal scope and terminologies. A system is built to identify similar cases based on the machine-readable, framebased knowledge representations of the judgments/documents. In this research, 4,835 US TM legal cases litigated in the US district and federal courts are collected as the experimental dataset. The computer-assisted system is constructed to extract critical features based on the ontology schema. The recommender will identify similar prior cases according to the values of their features embedded in these legal documents which include the case facts, issues under disputes, judgment holdings, and applicable rules and laws. Term frequency-inverse document frequency is used for text mining to discover the critical features of the litigated cases. Soft clustering algorithm, e.g., Latent Dirichlet Allocation, is applied to generate topics and the cases belonging to these topics. Thus, similar cases under each topic are identified for references. Through the analysis of the similarity between the cases based on the TM legal semantic analysis, the intelligent recommender provides precedents to support TM legal action and strategic planning.