Ontology-Based Legal Information Extraction 
Advised by: Amal Shehan Perera, Nisansa de Silva
University of Moratuwa
Reading and understanding a given piece of text is merely a walk in the park for a human being, if he or she is aware of the context of the text. This understanding process is supported by past experiences, domain knowledge and the structure of the given piece of text. With this understanding, a person could draw related conclusions to support the problem at hand. But this process is not as easy for computers, as it appears to be for humans. Drawing the conclusion to a given problem using the information and knowledge at hand, has always been a problem, in terms of efficiency, accuracy, and reliability. The information for a given domain increases day by day, making the entire process even more complex. Due to this reason, scientists have tried to come up with methods to get the necessary information as required from a given source, and this process is known as information extraction.
In Artificial Intelligence, information extraction is the drawing out of certain types of information from mostly natural language text, by processing them automatically [1]. Approaches like linguistic rules, gazetteer lists and web based search, are being used to carry out information extraction. Information extraction lies in between text understanding systems that attempt to analyze text and extract their semantic contents, and information retrieval systems which is merely a retrieval of information and documents related to the user's requirements [1]. In contrast, information extraction extracts information from natural language texts.
Given a domain of interest, it is necessary to generate an ontology or a knowledge base that would support the purpose of information extraction. "An ontology is an explicit specification of a conceptualization of a given domain" [2]. The use of ontologies for information extraction is known as ontology based information extraction. These systems may construct ontologies (by identifying concepts and relationships) and carries out the information extraction from necessary sources. An ontology is a meta-level description of the knowledge base of the domain of interest, where the ontology becomes a reusable component across that domain.
In the case of legal domains, "there is a massive legal literature in printed documents as well as online, where the current accessibility of sources of law does not suffice to serve legal and non-legal professionals" [3]. Implementing a reliable system that would be able to extract legal information from these online sources automatically, would be beneficial to legal officers and the general public as well. On the contrary, unlike in other domains, some of the legal case structures differ significantly [4], where special attention is needed to be given to the legal domain in the process of information extraction. One approach to solve this problem would be the use of ontologies in the legal domain. The generation of a legal ontology is not an easy task due to the various structures of cases and norms available in the domain. This difficulty can be mitigated by integrating domain knowledge into the ontology generation process. Afterward, the generated ontology can be used to guide the extraction of necessary information through an ontology based information extraction process. Having a user query system integrated with natural language processing to carry out the legal information extraction process, where the user simply has to enter the problem in natural text, would be an ideal solution to enhance the ontology based legal information extraction systems in the legal domain to provide easy interoperability and accessibility
Keywords: Natural Language Processing | Machine Learning / Deep Learning | Ontologies | Law | Big Data |