Search

This paper describes a long-term research goal which aims at creating a middleware interface between Argumentation Schemes and natural language. This idea comes from the need to face some challenges related to the automatic extraction of Argumentation Schemes from Nat- ural Language: for example the ability to extract Argumentation Schemes at different level of granularity. In the paper we describe how this process can be designed and how the structures of Argumentation Schemes can be modeled to this aim.

Privacy policies are known to be impenetrable and lengthy texts that are hardly read and poorly understood. This is why the General Data Protection Regulation (GDPR) introduces provisions to enhance information transparency including icons as visual means to clarify data practices. However, the research on the creation and evaluation of graphical symbols for the communication of legal concepts, which are generally abstract and unfamiliar to laypeople, is still in its infancy. Moreover, detailed visual representations can support users’ comprehension of the underlying concepts, but at the expense of simplicity and usability. This Chapter describes a methodology for the creation and evaluation of DaPIS, a machine-readable Data Protection Icon Set that was designed following human-centered methods drawn from the emerging discipline of Legal Design. Participatory design methods have ensured that the perspectives of legal experts, designers and other relevant stake- holders are combined in a fruitful dialogue, while user studies have empirically determined strengths and weaknesses of the icon set as communicative means for the legal sphere. Inputs from other disciplines were also fundamental: canonical principles drawn from aesthetics, ergonomics and semiotics were included in the methodology. Moreover, DaPIS is modeled on PrOnto, an ontology of the GDPR, thus offering a comprehensive solution for the Semantic Web. In combination with the description of a privacy policy in the legal standard XML Akoma Ntoso, such an approach makes the icons machine-readable and automatically retrievable. Icons can thus serve as information markers in lengthy privacy statements and support an efficient navigation of the document. In this way, different representations of legal information can be mapped and connected to enhance its comprehensibility: the lawyer-readable, the machine-readable, and the human-readable layers.

Under the General Data Protection Regulation (GDPR), transparency of information becomes an obligation aimed at creating an ecosystem where data subjects understand and control what happens to their personal data. The definition of transparency stresses its user-centric nature, while design considerations to comply with this obligation assume central importance. This article focuses on the icons established by the GDPR Art. 12.7 to offer “a meaningful overview of the intended processing”. Existing attempts to represent data protection through icons have not met widespread adoption and reasons about the strengths and weaknesses of their creation and evaluation are here discussed. Building on this analysis, we present an empirical research proposing a new icon set that responds to GDPR requirements. The article also discusses the challenges of creating and evaluating such icon set and provides some future directions of research for effective an effective implementation and standardization.

This work investigates how the combination of different tree representations with different Tree Kernel functions influences the results of the classifications in two specific case studies. One case study is related to the classification of argumentative stances of support, the other one is related to the classification of stances of opposition. Results show that some Tree Kernels achieves not only higher results but also a higher level of generalization. Moreover, it seems that also the kind of tree representation influences the performances of classifiers. In this study, we thus explore this relation between tree representation and different Tree Kernels, considering also compositional trees.

This Thesis is composed of a selection of studies realized between 2019 and 2022, whose aim is to find working methodologies of Artificial Intelligence (AI) and Machine Learning for the detection and classification of patterns and rules in argumentative and legal texts. We define our approach as “hybrid”, since different methods have been employed combining symbolic AI (which involves “top-dow” structured knowledge) and sub-symbolic AI (which involves “bottom-up” data-driven knowledge). The first group of these works was dedicated to the classification of argumentative patterns. Following the Waltonian model of argument (according to which arguments are composed by a set of premises and a conclusion), and the theory of Argumentation Schemes, this group of studies was focused on the detection of argumentative evidences of support and opposition. More precisely, the aim of these first works was to show that argumentative patterns of opposition and support could be classified at fine-grained levels and without resorting to highly engineered features. To show this, we firstly employed methodologies based on Tree Kernel classifiers and TFIDF. In these experiments, we explored different combinations of Tree Kernel calculation and different data structures (i.e., different tree structures). Also, some of these combinations employs a hybrid approach where the calculation of similarity among trees is influenced not only by the tree structures but also by a semantic layer (e.g. those using “smoothed” trees and “compositional” trees). After the encouraging results of this first phase, we explored the use of a new methodology which was deeply changing the NLP landscape exactly in that year, fostered and promoted by actors like Google, i.e. Transfer Learning and the use of language models. These newcomer methodologies markedly improved our previous results and provided us with stronger NLP tools. Using Transfer Learning, we were also able to perform a Sequence Labelling task for the recognition of the exact span of argumentative components (i.e. claims and premises), which is crucial to connect the sphere of natural language to the sphere of logic. The last part of this work was finally dedicated to show how to use Transfer Learning for the detection of rules and deontic modalities. In this case, we tried to explore a hybrid approach which combines structured knowledge coming from two LegalXML formats (i.e., Akoma Ntoso and LegalRuleML) with sub-symbolic knowledge coming from pre-trained (and then fine-tuned) neural architectures.

Changing or creating a new organization means creating a new process. Each process involves many risks that need to be identified and managed. The main risks considered here are procedural risks and legal risks. The former are related to the risks of errors that may occur during processes, while the latter are related to the compliance of processes with regulations. Therefore, managing the risks implies proposing changes to the processes that allow the desired result: an optimized process. In order to manage a company and optimize it in the best possible way, not only should the organizational aspect, risk management and legal compliance be taken into account, but it is important that they are all analyzed simultaneously with the aim of finding the right balance that satisfies them all. This is exactly the aim of this thesis, to provide methods and tools to balance these three characteristics, and to enable this type of optimization, ICT support is used. This work is not intended to be a computer science or law thesis but an interdisciplinary thesis. Most of the work done so far is vertical and in a specific domain. The particularity and aim of this thesis is not so much to carry out an in-depth analysis of a particular aspect, but rather to combine several important aspects, normally analyzed separately, which however have an impact on each other and influence each other. In order to carry out this kind of interdisciplinary analysis, the knowledge base of both areas was involved and the combination and collaboration of different experts in the various fields was necessary. Although the methodology described is generic and can be applied to all sectors, a particular use case was chosen to show its application. The case study considered is a new type of healthcare service that allows patients in acute disease to be hospitalized to their home. This provide the possibility to perform experiments using real hospital database.

Privacy policies are known to be impenetrable, lengthy, tedious texts that are hardly read and poorly understood. Therefore, the General Data Protection Regulation (GDPR) introduces provisions to enhance the transparency of such documents and suggests icons as visual elements to provide “in an easily visible, intelligible and clearly legible manner a meaningful overview of the intended processing.” The present dissertation discusses how design, and in particular legal design, can support the concrete implementation of the GDPR’s transparency obligation. Notwithstanding the many benefits that visual communication demonstrably provides, graphical elements do not improve comprehension per se. Research on graphical symbols for legal concepts is still scarce, while both the creation and consequent evaluation of icons depicting abstract or unfamiliar concepts represent a challenge. More- over, precision of representation can support the individuals’ sense-making of the meaning of graphical symbols, but at the expense of simplicity and us- ability. Hence, this research proposed a methodology that combines semantic web technologies with principles of semiotics and ergonomics, and empirical methods drawn from the emerging discipline of legal design, that was used to create and evaluate DaPIS, the Data Protection Icon Set meant to support the data subjects’ navigation of privacy policies. The icon set is modeled on PrOnto, an ontological representation of the GDPR, and is organized around its core modules: personal data, roles and agents, processing operations, processing purposes, legal bases, and data subjects’ rights. In combination with the description of a privacy policy in the legal standard XML Akoma Ntoso, such an approach makes the icons machine-readable and semi-automatically retrievable. Icons can thus serve as information markers in lengthy privacy statements and support the navigation of the text by the data subject.

One of the common characteristics of legal documents is the absolute preponderance of text and their specific domain language, whose complexity can result in impenetrability for those that have no legal expertise. In some experiments, visual communication has been introduced in legal documents to make their meaning clearer and more intelligible, whilst visualizations have also been automatically generated from semantically-enriched legal data. As part of an ongoing research that aims to create user-friendly privacy terms by integrating graphical elements and Semantic Web technologies, the process of creation and interpretation of visual legal concepts will be discussed. The analysis of current approaches to this subject represents the point of departure to propose an empirical methodology that is inspired by interaction and human-centered design practices.

The goal of this thesis is to present a multifaceted way of inducing semantic representation from legal documents as well as accessing information in a precise and timely manner. The thesis explored approaches for semantic information retrieval (IR) in the Legal context with a technique that maps specific parts of a text to the relevant concept. This technique relies on text segments, using the Latent Dirichlet Allocation (LDA), a topic modeling algorithm for performing text segmentation, expanding the concept using some Natural Language Processing techniques, and then associating the text segments to the concepts using a semi-supervised text similarity technique. This solves two problems, i.e., that of user specificity in formulating query, and information overload, for querying a large document collection with a set of concepts is more fine-grained since specific information, rather than full documents is retrieved. The second part of the thesis describes our Neural Network Relevance Model for E-Discovery Information Retrieval. Our algorithm is essentially a feature-rich Ensemble system with different component Neural Networks extracting different relevance signal. This model has been trained and evaluated on the TREC Legal track 2010 data. The performance of our models across board proves that it capture the semantics and relatedness between query and document which is important to the Legal Information Retrieval domain.

Companies must be able to demonstrate that their way of doing business is compliant with relevant rules and regulations. However, the law often has open texture; it is generic and needs to be interpreted before it can be applied in a specific case. Entrepreneurs generally lack the expertise to engage in the regulatory conversations that make up this interpretation process. In particular, for the application domain of technological startups, this leads to legal risks. This research seeks to develop a robust module for legal interpretation. We apply informal logic to bridge the gap between the principles of interpretation in legal theory with the legal rules that deter- mine the compliance of business processes. Accordingly, interpretive argu- ments characterized by argument schemes are applied to business models represented by value modeling (VDML). The specific outcome of the argu- mentation process (if any) is then summarized into a compliance pattern, in a context-problem-solution format. Two case studies in the application area of startups shows that the approach is able to express the legal arguments, but is also understandable for the target audience. The project is presented in two parts; Part I, the background, contains an introduction, literature review, motivational case studies, a survey on legal risks, and a modeling of business and legal aspects. Part II builds on the interdisciplinary facets of the first part to develop the Compliance Patterns Framework which is then validated with two case studies followed by a conclusion.

This thesis seeks to address the problem of the 'resource consumption bottleneck' of creating (legal) semantic technologies manually. It builds on research in legal theory, ontologies and natural language processing in order to semi-automatically normalise legislative text, extract definitions and structured norms, and link normative provisions to recitals. The output is intended to help make laws more accessible, understandable, and searchable in a legal document management system. Key contributions are: - an analysis of legislation and structured norms in legal ontologies and compliance systems in order to determine the kind of information that individuals and organisations require from legislation to understand their rights and duties; - an analysis of the semantic and structural challenges of legislative text for machine understanding; - a rule-based normalisation module to transform legislative text into regular sentences to facilitate natural language processing; - a Semantic Role Labeling based information extraction module to extract definitions and norms from legislation and represent them as structured norms in legal ontologies; - an analysis of the impact of recitals on the interpretation of legislative norms; - a Cosine Similarity based text similarity module to link recitals to relevant normative provisions; - a description of important challenges that have emerged from this research which may prove useful for future work in the extraction and linking of information from legislative text.

The objectives of the thesis are – (a) to identify the problems in water-energy-food nexus from ICT and Law point of view and to propose theoretically a legal knowledge framework for water-energy-food nexus in order to reduce those problems technologically, (b) to construct and implement legal ontology for nexus extracted from EU water, energy and food Regulations in OWL 2 language, which is a part of the grater work of implementing legal knowledge framework for water-energy-food nexus proposed through the compilation of objective (a). Considering these objectives, this thesis presents total five chapters. First chapter investigates current start of art of nexus in order to identify (1) major knowledge gaps in the nexus and (2) ontological existence of the nexus in the EU regulations, particularly in the legal definitions accommodated in EU Regulations and Directives associated with nexus domains. It also rationalizes the need for legal ontology for nexus. Second chapter evaluates existing perspectives and methodologies available for constructing legal ontology. The purpose of such evaluation was to select correct perspective and methodology for constructing legal ontology for nexus. It, at the end, justifies the need for developing new methodology for constructing the legal ontology for nexus. Third chapter explains the methodology used for engineering legal definitional knowledge extracted from the selected EU regulations in order to construct the legal ontology for nexus. Fourth chapter explains in detail the legal ontology for nexus while fifth chapter evaluates legal ontology for nexus. In addition, conclusion of the thesis shares critical issues faced throughout this doctoral thesis work. Furthermore, annexes contain a list of all formulas of restrictions implemented in legal ontology for nexus and links of all modules of legal ontology for nexus. LODE documentation of the legal ontology for nexus is available at http://codexml.cirsfid.unibo.it/post-doctoralresearchers/mizanur-rahman/.