An Annotation Language for Semantic Search of Legal Sources

An Annotation Language for Semantic Search of Legal Sources Adeline Nazarenko1, Franc¸ois Levy1, Adam Wyner2 1LIPN, Paris 13 University – Sorbonne Paris Cit´e & CNRS {Adeline.Nazarenko, Francois.Levy}@lipn.univ-paris13.fr 2Department of Computing Science, University of Aberdeen azwyner@abdn.ac.uk Abstract While formalizing legal sources is an important challenge, the generation of a formal representation from legal texts has been far less considered and requires considerable expertise. In order to improve the uniformity, richness, and efficiency of legal annotation, it is necessary to experiment with annotations and the annotation process. This paper reports on a first experiment, which was a campaign to annotate legal instruments provided by the Scottish Government’s Parliamentary Counsel Office and bearing on Scottish smoking legislation and regulation. A small set of elements related to LegalRuleML was used. An initial guideline manual was produced to annotate the text using annotations related to these elements. The resulting annotated corpus is converted into a LegalRuleML XML compliant document, then made available via an online visualisation and query tool. In the course of annotating the documents, a range of important interpretive and practical issues arose, highlighting the value of a focused study on legal text annotation. Keywords: annotation, semantic search, legal information, methodology 1. Introduction Formalizing legal sources has been identified for years as an important challenge for the development of legal content management in order to increase interoperability or support dialogue between various legal systems and actors. The generation of the envisaged formal representation from the legal texts has been far less considered and requires con- siderable expertise. Nevertheless, enriching the texts with a layer of semantic annotations and opening new access routes to textual content represent a critical issue for all the legal work that rests essentially on the analysis and interro- gation of sources. Legal annotation requires legal skills to understand the text and the significance of legal statements as well as logical skills to understand what conclusions can be drawn with re- spect to the annotations. In order to improve the uniformity and the cost of legal annotation, it is necessary to exper- iment with annotations which allow us to query, analyse, and interrogate legal sources. Our research was driven by requirements set by the parlia- mentary counsel of the Scottish Government’s Parliamen- tary Counsel Office, which aims to improve internal leg- islative drafting. As part of this, it is useful to provide a corpus of law in electronically readable form which can be queried to address the following competence questions: 1. What are all the offences and associated penalties or defences? 2. What prohibitions apply to tobacco products? 3. What obligations have been placed on which entities, e.g. shop owners? 4. What permissions are given to Scottish Ministers? 5. Given a provision, what are related overriding or repa- ration provisions? For parliamentary counsels, who draft the law, the answers to these questions ought to be all relevant provisions so that they can compare formulations or check interactions be- tween provisions. For the technical solution to extracting such questions, semantic meta-data …

Content truncated to 500 words