In light of the challenges in tracking and comprehending diverse interpretations of legal texts, this paper introduces a method for norm interpretation using ontologies and digital tools. We propose the FLINT ontology, a language for expressing interpretations of normative sources, as well as a set of software tools to support its utilization. The FLINT ontology represents norms as normative acts with pre- and postconditions. Each of these elements are linked to its source, ensuring traceability and explainability. The software tools we introduce in this paper include a source decomposition application that converts normative documents into RDF-based representations, a norm editor for creating interpretations of norms based on FLINT, and the integration of an automated assistance module that generates annotation suggestions to assist users in encoding normative texts.
Norms can be subjected to multiple interpretations. Not having a reference model allowing to indicate which interpretation has been applied to a given normative source, such as a legal text or regulation, makes it challenging to track from which perspective it has been interpreted. This issue becomes even more problematic when trying to keep track of decision-making processes, as the interpretation of rules directly influences the outcomes.
Ontologies can help to record information regarding the interpretation of norms. When they are integrated within digital tools, they facilitate normative cooperation and foster further reflection on how a certain regulation has been interpreted by diferent users. Moreover, they may allow stakeholders to reach deeper insights into the underlying logic of norms, leading them to further discuss on the interpretation that should be applied to norms.
To enable eficient tracking of the interpretation that has been employed to structure data
of rules and regulations, we worked on solutions fostering the representation of normative
knowledge based on a sound theoretical foundation. As part of this endeavour, we developed
an ontology for the FLINT language [
The rest of the paper is set up as follows. We start by giving a brief overview of the state of the art of normative ontologies and the tooling that is available to utilize them in Section 2. We then introduce the FLINT ontology in Section 3 and describe the tooling in Section 4. We conclude in Section 5.
Numerous vocabularies and ontologies have been developed to structure and organize data
relating to norms [
Arguably the most relevant comparison for our work is to the UFO-L Legal Core Ontology
[
A line of work focused on mapping normative texts to formal models by means of an
annotation tool is that of Dynamic Condition Response (DCR, [
FLINT is a language that can be used to express interpretations of normative sources, while also serving as a basis for technical implementation. It does so by representing norms in terms of normative acts and their pre- and postconditions. The FLINT ontology is developed in RDF/OWL and is publicly available under an open-source license.1
FLINT’s central concept is the frame: a container that bundles several pieces of information
[
An important characteristic of FLINT, which is reflected in our ontology, is that each frame contains a reference to the source that this frame is an interpretation of (e.g. one specific sentence in a law text). These references ensure that the interpretation is explainable and traceable. To support fine-grained referencing, we developed a source ontology module that describes the structure and content of a source of norms.
The software modules that we designed aim to support data-structuring activities related to norms while providing the possibility to everyone to specify and inspect information regarding rules and regulations. The target users for our tooling are legal experts, public administrators, non-profit organizations, businesses, and citizens. In fact, we envision a world where anyone can contribute to the clarification of the normative systems they participate in.
The tools are meant to exchange data in a standardized way, following the previously mentioned ontology modules. This ensures that the tools can also be used independently and in combination with other software. We describe the tools in detail in the following subsections.
The first module of our tooling is an application for preparing any source of norms into a format that allows fine-grained referencing. This application, which we named Choppr, converts a document into an RDF-based representation of the structure and contents of that document.2 The process is semi-automatic: our source decomposition tool takes as input any text file, and allows the user to develop and test a set of instructions for the decomposition of the document. The instructions are based on scoped, sequential application of regular expressions. The tool can export the decomposition result as RDF-compliant data, using the concepts of the source ontology module.
The second module of our tooling is the norm editor.3 It is an application, built using web-based technologies, which allows a user to create interpretations of sources of norms in FLINT in a user-friendly and interactive way. The code of the editor is open source.4
The editor consists of three main panes. In the source view pane, a decomposed source as it is generated by the Choppr tool can be imported. The decomposed source is rendered and allows the user to annotate pieces of textual information according to the expressiveness of the FLINT ontology (acts, facts and their subtypes). It provides the capability to continuously observe the relationship between each FLINT frame and the corresponding source of the regulation. In the 2A prototype, developed in collaboration with software company Tick, is available at https://choppr.app/. 3A prototype of the norm editor can be accessed at https://norm-editor.tnodatalab.nl/. 4The repository can be accessed at https://gitlab.com/normativesystems/ui/interpretation-editor. interpretation actions pane, an overview of the already annotated parts is given. Additionally, the user can select an annotation and include it in the supported interpretations actions (creation of facts and acts) of the right pane. The final step, exporting the interpretation in RDF, is under development.
In the process of encoding textual information into the FLINT ontology, users are supported
by automated suggestions. For this we use a software module that can automatically generate
tag suggestions for stakeholders, developed by Bakker et al. [
The information exchanged between the components of our software tooling follows the path that users undertake to structure normative textual information. These steps include decomposing a source, manually or automatically encoding the segmented texts into the expressiveness of the FLINT ontology, and saving FLINT encoded information into the repository of choice. By following this process, the generation of structured data regarding norms becomes a streamlined process, allowing for easy sharing and dissemination of information. It ultimately supports the creation of machine-understandable data that can feed diferent sectoral ecosystems while contributing to interoperability of information.
In this paper, we reported on our developments of tooling for legal interpretation. We presented OWL ontologies formalizing relevant concepts, including a language to describe normative systems, and a representation of the content and structure of normative sources. We presented software tooling to support decomposition of normative sources, and manual and computerassisted interpretation of those sources.
The tooling presented in this paper is still in active development. Usability tests are planned in the fall of 2023. Although FLINT has been tested in case studies, we have so far worked with manually created models without the use of tooling. We are now working on case studies to also test the maturity of the tools. For example, we are creating an interpretation of the Dutch budget laws in order to see whether we can accurately model the national budget cycle.
In addition, we are pursuing computational implementations of FLINT interpretations in order to automate normative reasoning. A norm engine based on the FLINT ontology is under development, which uses the ontology in combination with SHACL constraints and inference rules to reason about the lawfulness of acts in concrete scenarios.
As part of our future work, we intend to conduct a comprehensive comparison with the other normative ontologies and tooling mentioned in this paper, aiming to gain valuable insights into their similarities, diferences, and overall eficacy.
We thank the Dutch Ministry of the Interior and Kingdom Relations for financially supporting our research. We are also grateful to Kamiel Naaktgeboren and Tinus Koorn for the development of Choppr and to Rutger Dik and Robert van Doesburg for their work on Choppr and the norm editor.