Author:

Abstract:

The huge amount of documents available in the legal domain calls for computational tools supporting efficient and intelligent search and filtering of information. Over the last several years, machine-learning oriented research in information retrieval and document classification has spawned a number of systems capable of handling structural content management, helping users to automatically or semi-automatically identify relevant structured portions of legal texts, such as paragraphs, chapters or intertextual references.This PhD thesis explores a novel idea to identify relevant portions of legal texts by using argumentation analysis. Many legal texts are argumentative, such as the case files exchanged by the parties in a case, a court's decision, scholarly publications and discussions and opinions in legal blogs. Therefore, argumentation can be used as a means to structure the texts contents to search and filter their information. However, there has been little research done on the automatic detection of argumentation or its structure.This thesis presents a general way to automatically detect the arguments of a legal text and how they interact. This allows one to obtain a structured representation of the information of the text which later on can be used as a novel means to search or filter documents.To this end this thesis introduces and discusses the development of the first corpus of legal texts fully annotated by their argumentation, including the four stages of the corpus creation process (design, collection, annotation and analysis). It also presents different approaches to obtain an automatic method to detect argumentation in legal cases. All the approaches are based on state-of-the-art information retrieval and natural language processing methods.