Ontological solution for Sri Lankan legal system

Abstract

Body of knowledge in legal domain is complex, unstructured, interconnected in many ways and constantly being updated. This has resulted in hindering general public to access the legal domain, and creating gap between society and the legal system. Legal professionals also find it difficult to refer to case with adequate depth and the breadth. This research has been conducted to develop Ontology for legal domain with a particular emphasis on constitutional law in Sri Lanka. The research has focused on 1978 constitute including the 19th amendment which was passed in 2015 May. Semi-Automated mechanism was used to construct ontology. After ontology is constructed, subject matter expert has corrected the Ontology. End users has the option of querying the Ontology for simple text as that is the current procedure in the legal domain. End users have the option of execute queries against the ontology. Protégé has been selected as the development environment for propose ontological system. To enhance performance of the system, relational database management system was used to implement Ontology. This system can be used by Lawyers and students by running queries to get relevant answers to their questions. In addition to the answers to the question, this system will provide the appropriate legal act and any other relevant legal cases. Natural Language Processing and Text Mining techniques were used to identify user queries. Ontology has questions and user questions are matched against those questions. Each question is mapped to one or many incidents, content and cases. Also, Agent technology is used to extract updated legal documents content and updated to legal Ontology. To enhance the performance of the knowledge base, relational database management system was used. Two verifications were used. First, verify whether users question and knowledge base questions are correctly matched. Secondly, time taken to answer and correctiveness are taken as the parameters. When all the scenarios are considered ranking relevancy is at least than 60% percent and in some scenarios it is 100 percent.