The automatic detection of arguments in text regards a relatively new area at the intersection of Natural Language Processing, Information Retrieval and Legal Information Systems. This paper presents some fundamental issues when processing texts that contain argumentation. Furthermore, our research bridges different areas, including the legal field and the Semantic Web, where argumentation detection and reconstruction could be beneficial. Finally, it analyses several methodologies to accomplish this task, providing results from different experiments done over several kinds of texts, specially legal reports.