Managing semantics in XML vocabularies: an experience in the legal and legislative domain

Abstract

Akoma Ntoso is an XML vocabulary for legal and legislative documents sponsored by the United Nations, initially for African Countries and subsequently for use in other world countries. The XML documents that represent legal and legislative resources in Akoma Ntoso contain a large quantity of elements and sections with concrete semantic information about the correct description and identification of the resource itself and the legal knowledge it contains. Such information is organized in many distinct conceptual layers, allowing for the contribution of different semantic information according to competencies and role in the workflow of the contributor. This paper shows how the Akoma Ntoso standard expresses the independent conceptual layers of semantic information, and provides ontological structures on top of them. We also discuss how current Semantic Web technologies could be used on these layers to reason on the underlying legal texts. As one of the main funding principles of Akoma Ntoso is the long-term preservation of legal documents and of their intended meaning, this paper also shows and justifies some design decisions that have been made in order allow future toolmakers to access the enclosed legal information without having to rely on current technology that may be long forgotten in the future decades.

Deep learning based multi-label text classification of UNGA resolutions

“…Each keyword is also connected with a Top Level Class (TLC) Concept using the @refersTo attribute. The TLCs are basic classes of the non-ontology of Akoma Ntoso [29][30] that permits to connect each fragment of the document to the real ontology (in this case the SDGIO ).…”

Hybrid Refining Approach of PrOnto Ontology

“…Let the GDPR and the Privacy Policies be our corpus C. In order to perform the automatic text annotation of our corpus with PrOnto concepts, we follow these steps: 1. we firstly identify a list of all the terms (subjects, objects, verbs) in C, by using a simple variant of ClauseIE. The identified terms are said to be possible classes (in the case of subjects and objects) or possible properties (in the case of verbs); 2. we use PrOnto labels of classes and properties, with additional mapping of linguistic and lexical variants; 3. we try to map every possible class/property in C to its closest class/property in PrOnto, by using the same algorithm used in a previous project 4 . This algorithm exploits pretrained linguistic deep models in order to be able to easily compute a similarity score between two terms.…”

Legislative Drafting Systems

“…Most of these applications need to address the difficult balance of three main aspects (Barabucci et al, 2010):…”