In machine learning, splitting data into training and test sets is usually achieved using random stratified sampling, in which classes are proportionally divided into two subsets. Other methods also consider feature-aware criteria, and some of those methods claim to have achieved optimal split of minimised variance. We do not advocate aiming to achieve an optimal split or minimise variance since this would be counterproductive for ensemble methods, where the diversity of the training set is desired. Ensemble methods achieve diversity through bagging and boosting schemes. In the recently introduced Monte Carlo ensemble approach, diversity can be maintained through random stratified sampling without using bagging or boosting methods. This work introduces a feature-aware split that retains the diversity of the ensemble. To this end, we propose an evolutionary algorithm that starts with an entirely random population and aims at objectives of proportional class-representation and minimisation of the normalized mean error rather than minimisation of variance. The proposed data-split method is tested on three different models used within a hierarchical Monte Carlo ensemble. The results show that the method positively affects the predictability performance when applied on two domain-specific material science datasets and a collection of 38 general machine learning datasets.

The task of forecasting case importance has received far less attention in the legal domain compared to judgment prediction, but it is a task that is an essential step to capture within tools for assisting with the processing of cases submitted to a court. In this paper we propose a cornerstone framework for carrying out the task of forecasting case importance, using communicated cases, which are documents available prior to any decision being issued. The setting for our work is cases in the European Court of Human Rights, with a specific focus on Article 3, prohibition of torture. We set out proposals for a Retrieval-Augmented Generation (RAG) framework that makes use of Large Language Models augmented with Semantic Search, Knowledge Graph and Re-Ranking components and we evaluate the effectiveness of this framework and its components. Further experiments conducted evaluate the framework using both pre-trained and fine-tuned LLMs, as well as use of different prompting strategies. The basic experiments show a propensity for the LLMs to significantly overestimate the importance of cases, but when we augment the LLMs with the aforementioned components, we are able to gain uplifts in performance. Our framework and results provide a solid basis for determining the requirements for the development of successful automated tools to be used to assist with determining case importance.

We report on a first set of results from experiments undertaken to tackle the novel task of determining the optimal context window for extracting and contextualising citation instances within case law. The wider task of outcome prediction using AI tools cannot be undertaken without considering the role that citations play when new cases are being decided. This short paper aims to shine a light on the importance of this task and provide the foundation for developing AI tools to capture citations’ context by examining a corpus of legal cases taken from the European Court of Human Rights. Our results show that there is an identifiable “Goldilocks Zone” of scoped paragraph-level context windows that attention can be focused on for extracting citation instances.

We present a context-aware approach to citation analysis for European Court of Human Rights case law. Instead of a single edge between cases, we annotate each citation instance with respect to its individual context. We construct two datasets: a new human-annotated set spanning judgments and decisions across Grand Chamber, Chamber, and Committee, and an AI-annotated set of judgments produced at scale. We find that most authorities are cited at least twice within a case, and repeated citations frequently vary by complaint and judicial consideration. Our empirical analysis shows that AI annotations are highly competitive with outputs from trained human annotators, but with significant variation depending on the provision of convention and the level of the Court producing the ruling. We release the datasets and an interactive Citation Analysis Tool that enables context-filtered retrieval, supporting triage and research grounded in past precedent.

Online reviews now have a considerable influence on consumer choices. However, little work has focused on what features of review platforms influence review quality. We present a novel approach to identify the features that encourage quality reviews. By interpreting reviews as arguments for or against the product, an argument scheme can be used to simulate the emergent reliability of reviews resulting from different setups of the online review platform. Our results show that if the most recent, helpful, or polarised reviews are promoted over quality, then good quality reviews will almost never be shown to users.

This demo paper introduces an application that is capable of identifying and visualising diagnostic arguments within abstract argumentation systems. The software presented is underpinned by a novel algorithm, called the Diagnostic Argument Identifier, that combines a semantic-based approach with a technique from the information-theoretic literature, to quantify the impact that the removal of an argument has on the acceptability of other arguments.

This paper presents the theory of probabilistic hier-archical supervised ensemble learning (TPHSEL), a classification approach we have developed with the goal of obtaining classifications for material selection with a degree of interpretability of the results. We found that TPHSEL is a competitive classifier, not only for our target application, but also for a broader range of standard datasets, where it outperformed support vector machines, random forests, and optimal classification trees. The dataset we developed the method for within the field of materials science is small (405 entries), leading to relatively low accuracy (81 % to 82 %) for both our method and a deep learning approach used earlier. In this context, we found that selection based on a large vote share left close to 20 % of candidate materials, and in this bracket, accuracy and other model performance metrics are above 0.95. This is excellent news for prioritising experimental targets (and related tasks), as it indicates that it is possible to identify promising candidates based on data that still leaves shortfalls in classification.

This paper presents hierarchical supervised Monte Carlo ensemble learning (HSMEL). This provides an extension to the theory of probabilistic hierarchical supervised ensemble learning (TPHSEL), which itself evolved from the theory of prob-abilistic hierarchical supervised learning (TPHSL). The basic idea captured in TPHSL is that a complex model can be replaced with a hierarchy of simple and mathematically understandable models. Such models are amenable to interpretation, and they are therefore more likely to contribute to explainable AI, in comparison to black box models. The basic TPHSL was subsequently advanced to TPHSEL, where several hierarchical models make a classification decision by majority vote. In this paper TPHSEL is further advanced to include the notion of Monte Carlo ensemble. We show that this ensemble is computationally faster and has broader reach on training examples. The method has been deployed in use cases from materials science, specifically to study the impact of various features on the conductivity of materials. Based on the performance of individual features, the method has been devised that applies set theory over ensemble outcomes to predict the average accuracy that could be achieved if those features are grouped in some way. We argue that this method has potential to accelerate material design procedures by providing predictions about machine learning performance parameters without engaging in extensive computational effort and consequently will also reduce chemistry lab experimentation. In addition, to show resilience of HSMEL, we have also applied it on 28 general machine learning datasets, where its performance is compared with the classical methods from the literature.

With the exponential growth of digital information, ensuring text security, a fundamental component of information security, becomes increasingly paramount. While authentication remains a primary focus for data access control and protection, the rich sensor ecosystem and immersive experiences of virtual reality (VR) environments introduce new privacy risks, particularly with inconspicuous sensors like motion and location sensors. In this context, protecting the security of text entered by users poses a unique challenge. This paper explores the feasibility of enhancing text security by introducing variability in virtual input tools during typing processes. Specifically, we investigate the impact of introducing successive and random intermittent variations to the virtual ray (start point and direction) with controller-based raycasting techniques on text security and typing experience. The results demonstrate that introducing variability in virtual ray effectively protects regular text and passwords. Random intermittent introducing variability balances security and user experience for regular text. These findings provide insights into enhancing text security beyond authentication and defending against the potential risks in VR environments.

We report on a study undertaken to analyse AI performance on two tasks involved in automating processing of cases from the European Court of Human Rights: classification of legal case outcomes and keyword prediction. Results show variation across Articles and Court levels, and challenge the common viewpoint that larger legal corpora combined with larger models will be sufficient for effective automated legal reasoning. Legal summarisation, as reflected with keyword prediction, proved more challenging than outcome classification. Our results suggest the need for improved case law retrieval and understanding of contextual factors for effective automated legal decision support.

Alphanumeric and special characters are essential during text entry. Text entry in virtual reality (VR) is usually performed on a virtual Qwerty keyboard to minimize the need to learn new layouts. As such, entering capitals, symbols, and numbers in VR is often a direct migration from a physical/touchscreen Qwerty keyboard-that is, using the mode-switching keys to switch between different types of characters and symbols. However, there are inherent differences between a keyboard in VR and a physical/touchscreen keyboard, and as such, a direct adaptation of mode-switching via switch keys may not be suitable for VR. The high flexibility afforded by VR opens up more possibilities for entering alphanumeric and special characters using the Qwerty layout. In this work, we designed two controller-based raycasting text entry methods for alphanumeric and special characters input (Layer-ButtonSwitch and Key-ButtonSwitch) and compared them with two other methods (Standard Qwerty Keyboard and Layer-PointSwitch) that were derived from physical and soft Qwerty keyboards. We explored the performance and user preference of these four methods via two user studies (one short-term and one prolonged use), where participants were instructed to input text containing alphanumeric and special characters. Our results show that Layer-ButtonSwitch led to the highest statistically significant performance, followed by Key-ButtonSwitch and Standard Qwerty Keyboard, while Layer-PointSwitch had the slowest speed. With continuous practice, participants' performance using Key-ButtonSwitch reached that of Layer-ButtonSwitch. Further, the results show that the key-level layout used in Key-ButtonSwitch led users to parallel mode switching and character input operations because this layout showed all characters on one layer. We distill three recommendations from the results that can help guide the design of text entry techniques for alphanumeric and special characters in VR.

The purpose of this paper is to provide a definitive, up-to-date account of a methodology has that been proven successful for representing and reasoning about legal domains. The ANGELIC (ADF for kNowledGe Encapsulation of Legal Information for Cases) methodology was originally developed to exploit then recent developments in knowledge representation techniques that lend themselves well to capturing factor-based reasoning about legal cases. The methodology is situated firmly within the tradition of research in AI and Law that aims to build systems that are knowledge rich in terms of the domain expertise that is emulated within the systems. When the methodology was first introduced, it was demonstrated on academic examples, but it was subsequently used in and evaluated on a variety of real world domains for external clients. This set of evaluation exercises yielded a variety of learning points as the methodology was applied to different legal domains with their own particular features. These learning points, and the extensions to the methodology that follow from them, urge a consolidation exercise to provide an updated version of the methodology that reflects how it has matured over time. This paper represents a milestone in the development of the methodology in that it presents the ANGELIC II Domain Model, along with a description of its constituent parts, and demonstrates its application through a case study in a key evaluation domain.

Recent years have witnessed significant progress in the deployment of advanced Natural Language Processing (NLP) techniques based on transformer technology, across many domains and applications. However, in legal domains, due to the complexity, length, and sparsity of legal case documents, the use of these advanced NLP techniques has offered comparatively slight returns. Perhaps even more importantly, such methods are critically lacking in explainability and justification of outputs, which are essential for many legal applications. We propose that the direction of these NLP techniques should be aimed at ascription to a legal knowledge model, which can then provide the necessary and auditable justifications for the rationale of any case outcome. In this paper we investigate the effectiveness of using Hierarchical Bidirectional Encoder Representations from Transformers (H-BERT) models to ascribe to an Angelic Domain Model (ADM) that is able to represent the legal knowledge of a domain in a structured way, enabling justifications and improving performance. Our study involved an annotation task on a popular domain, cases from the European Court of Human Rights, to gain an understanding of the balance of complaints in the domain. The data set produced from this study enabled training of models for factor ascription using the classification targets derived from the annotations. We present results of experiments conducted to evaluate the performance of the ascription task at three different levels of abstraction within the structured model.

Current theories of precedential constraint attempt to incorporate dimensions into the reasons for decisions. We argue that this is an unnecessary complication, and precedential constraint can be handled using only factors. In our account the role of dimensions is to organise facts, and their effect operates at the factor ascription level, prior to precedential constraint being applied.

We report a study undertaken to analyse human performance on the verdict classification task. Several approaches have addressed this task with outcomes compared against the outcomes from actual legal cases. Results vary and we investigate how classification is done by humans. A key finding is that fact descriptions alone are insufficient for accurate classification, independent of legal background.

In this paper we report on a significant research project undertaken to design, implement and evaluate explainable decision-support tools for deciding legal cases. We provide a model of a legal domain, Article 6 of the European Convention on Human Rights, constructed using a methodology from the field of computational models of argument. We describe how the formal model has been developed, extended and transformed into practical tools, which were then used in evaluation exercises to determine the effectiveness and usability of the tools. The underpinning AI techniques used yield a level of explanation that is firmly grounded in legal reasoning and is also digestible by the target end users, as demonstrated through our evaluation activities. The results of our experimental evaluation show that on the first pass, our tool achieved an accuracy rate of 97% in matching the actual decisions of the cases and the user studies conducted gave highly encouraging results with respect to usability. As such, our project demonstrates how trustworthy AI tools can be built for a real world legal domain where critical needs of the end users are accounted for.

Reasoning with legal cases by balancing factors (reasons to decide for and against the disputing parties) is a two stage process: first the factors must be ascribed and then these reasons for and against weighed to reach a decision. While the task of determining which set of reasons is stronger has received much attention, the task of factor ascription has not. Here we present a set of argument schemes for factor ascription, illustrated with a detailed example.

Reasoning with legal cases has long been modelled using symbolic methods. In recent years, the increased availability of legal data together with improved machine learning techniques has led to an explosion of interest in data-driven methods being applied to the problem of predicting outcomes of legal cases. Although encouraging results have been reported, they are unable to justify the outcomes produced in satisfactory legal terms and do not exploit the structure inherent within legal domains; in particular, with respect to the issues and factors relevant to the decision. In this paper we present the technical foundations of a novel hybrid approach to reasoning with legal cases, using Abstract Dialectical Frameworks (ADFs) in conjunction with hierarchical BERT. ADFs are used to represent the legal knowledge of a domain in a structured way to enable justifications and improve performance. The machine learning is targeted at the task of factor ascription; once factors present in a case are ascribed, the outcome follows from reasoning over the ADF. To realise this hybrid approach, we present a new hybrid system to enable factor ascription, envisioned for use in legal domains, such as the European Convention on Human Rights that is used frequently in modelling experiments.

One approach to building legal support systems is to run an executable model of the relevant knowledge through an interface designed to collect information from the user and provide explanations. The usability of such systems depends on the terms used in the law being represented: often only users familiar with the practice and application of the law will be able to provide the required information. Earlier work applied this approach to the European Convention on Human Rights (ECHR). Although the performance of the tool built for that domain was good, the questions posed to the user demanded a good deal of knowledge and experience of the ECHR. Here we use the knowledge of an expert with extensive experience of the ECHR to extend the model, through intermediate levels, to identify questions that are appropriate to the target user. We have undertaken a pilot evaluation in which a small number of lawyers have used the prototype program and provided very positive feedback, showing that they are receptive to AI solutions that give effective, explainable decision support.

Horty, Rigoni and Prakken have developed formal characterisations of precedential constraint based on dimensions and factors as introduced in HYPO and CATO. We discuss the relation between dimensions and factors and also describe the current models of precedential constraint based on factors, along with some criticisms of them. We argue that problems arise from ignoring the structure of legal cases that is provided by the notion of issues, and that seeing precedential constraint in terms of issues rather than whole cases provides a more effective approach and better reflects legal practice. The advantages of the issue based approach are illustrated with a concrete example. We then discuss how dimensions should be accommodated, suggesting that this is best done by seeing reasoning with legal cases as a two stage process: first factors are ascribed to cases and then factor based reasoning can be used to arrive at a decision. Thus precedential constraint can be described in terms of factors, dimensions being handled at the first stage. Both stages are constrained, in different ways, by precedents: we identify three types of precedent: framework precedents which structure cases into issues, preference precedents which resolve conflicts between opposing sets of factors within these issues, and ascription precedents which constrain the mapping from facts to factors.

Explanation and justification of legal decisions has become a highly relevant topic in light of the explosion of interest in the use of machine learning (ML) approaches to predict legal decisions. Current suggestions are to use the established factor based explanations developed in AI and Law as the basis for explaining such programs. We, however, identify factor ascription as an important aspect of explanation of case outcomes not currently covered, and argue that explanations must also include this aspect. Finally, we outline our proposal for a hybrid system approach that combines ML and Abstract Dialectical Framework (ADF) layers to engender an explainable process.

Abstract An abstract argumentation framework is a directed graph $(V,E)$ such that the vertices of $V$ denote abstract arguments and $E \subseteq V \times V$ represents the attack relation between them. We present a new ad hoc algorithm for computing the grounded extension of an abstract argumentation framework. We show that the new algorithm runs in $\mathcal{O}(|V|+|E|)$ time. In contrast, the existing state-of-the-art algorithm runs in $\mathcal{O}(|V|+|S||E|)$ time where $S$ is the grounded extension of the input graph.

In this paper we present an argumentation-based approach to representing and reasoning about a domain of law that has previously been addressed through a machine learning approach. The domain concerns cases that all fall within the remit of a specific Article within the European Court of Human Rights. We perform a comparison between the approaches, based on two criteria: ability of the model to accurately replicate the decision that was made in the real life legal cases within the particular domain, and the quality of the explanation provided by the models. Our initial results show that the system based on the argumentation approach improves on the machine learning results in terms of accuracy, and can explain its outcomes in terms of the issue on which the case turned, and the factors that were crucial in arriving at the conclusion.

In the context of abstract argumentation frameworks, the admissibility problem is about deciding whether a given argument (i.e. piece of knowledge) is admissible in a conflicting knowledge base. In this paper we present an enhanced backtracking-based algorithm for solving the admissibility problem. The algorithm performs successfully when applied to a wide range of benchmark abstract argumentation frameworks and when compared to the state-of-the-art algorithm.

Modelling reasoning with precedents has been a central concern of AI and Law since its inception. A recent paper has provided a discussion (in jurisprudential terms) of whether such reasoning is best seen as rule application or analogy. We review some of the prominent AI and Law approaches and find that over the years there has been a move away from analogy to rule application. Even in those approaches which do use analogy, however, the analogies handled concern only analogies between cases represented as sets of factors, and do not consider analogies between the elements of the fact situations peculiar to particular cases. In actual practice, however, analogies are used to determine which factors are relevant in a case, and which party is favoured by particular aspects of the case situation. Such analogies relate not to factors, but to real-world elements of the case and are hard to make and critique without a comprehensive common sense ontology. Thus while we may be able to construct specific ontologies to model past examples of such analogical reasoning, which can be useful for simulation and teaching, the ability to perform analogical reasoning on novel situations is, and is likely to remain, infeasible. This conclusion suggests that there will always be limits to our ability to construct systems able to handle new cases presenting novel situations.

Coherent division of legal document bundles, whether this is done in the context of court bundles, briefs or some other application, is a time consuming and challenging task. We propose an approach whereby this process can be automated. Two variations are considered. The first addresses the scenario where the topic labelling is pre-defined and adopts a supervised learning approach. The second addresses the scenario where the topic labelling, for whatever reason, is not specified in advance and adopts an unsupervised learning approach. This paper reports on an investigation of both mechanisms using accident claims bundles. The evaluation results indicate that the proposed approaches can be successfully applied to divide legal document bundles.

© 2019 The authors and IOS Press. ANGELIC is a methodology for encapsulating knowledge of a body of case law. Logiak is a system intended to support the development of logic programs by domain experts, and provides an excellent environment for the rapid realisation of ANGELIC designs. We report our use of Logiak to realise ANGELIC designs, using both Boolean factors and factors with magnitude.

Lifelong Machine Learning (LML) mimics common human learning experiences. Humans undergo through long learning phase at start while studying followed by updating knowledge base incrementally from everyday instances. The objective is to retain past learnt knowledge and transfer learning to the next task iteratively. Training on the large data pool through a one-shot long running batch job limits the responsiveness and increases the infrastructure cost through large cluster requirements. The full dataset may not be available as well at the initiation of the training process. Through a review of previous work on lifelong machine leaning, we propose a Multi-agent Lambda Architecture (MALA) model to combine historical batch data with live streaming data to develop a lifelong learning system. MALA allows the streaming process to initialize itself with trained model from the batch data. Streaming process takes the batch data offset and incrementally updates the model iteratively with new waves of data. Reasons for our claim are presented through implementation of a recommender engine.

We relate the ANGELIC methodology for acquiring and encapsulating domain knowledge to the ASPIC+ framework for structured argumentation. In so doing we hope to facilitate the building of applications in concrete domains by linking a successful methodology to a proven theoretical framework. We use an example from the ASPIC+ literature to illustrate the relationship.

We discuss the lessons learned from implementing a CATO style system using factors with magnitude. In particular we identify that giving factors magnitudes enables a diversity of reasoning styles and arguments. We distinguish a variety of ways in which factors combine to determine abstract factors. We discuss several different roles for values. Finally we identify the additional value related information required to produce a working program: thresholds and weights as well as a simple preference ordering.

The decision whether to accept or reject a new case is a well established task undertaken in legal work. This task frequently necessitates domain knowledge and is consequently resource expensive. In this paper it is proposed that early rejection/acceptance of at least a proportion of new cases can be effectively achieved without requiring significant human intervention. The paper proposes, and evaluates, five different AI techniques whereby early case reject-accept can be achieved. The results suggest it is possible for at least a proportion of cases to be processed in this way.

A development environment for the Angelic Methodology.

The effectiveness of document Information Extraction (IE) is greatly affected by the structure and layout of the documents being considered. In the case of legal documents relating to commercial law, an additional challenge is the many different and varied formats, structures and layouts used. In this paper, we present work on a flexible and scalable IE environment, the CLIEL (Commercial Law Information Extraction based on Layout) environment, for application to commercial law documentation that allows layout rules to be derived and then utilised to support IE. The proposed CLIEL environment operates using NLP (Natural Language Processing) techniques, JAPE (Java Annotation Patterns Engine) rules and some GATE (General Architecture for Text Engineering) modules. The system is fully described and evaluated using a commercial law document corpus. The results demonstrate that considering the layout is beneficial for extracting data point instances from legal document collections.

We describe the use of the ANGELIC methodology, developed to encapsulate knowledge of particular legal domains, to build a full scale practical application for internal use by a firm of legal practitioners. We describe the application, the sources used, the stages in development and the application. Some evaluation of the project and its potential for further development is given. The project represents an important step in demonstrating that academic research can prove useful to legal practitioners confronted by real legal tasks.

We build on two recent attempts to formalise reasoning with dimensions which effectively map dimensions into factors. These enable propositional reasoning, but sometimes a balance between dimensions needs to be struck, and to permit trade offs we need to keep the magnitudes and so reason more geometrically. We discuss dimensions and values, arguing that values can play several distinct roles, both explaining preferences between factors and indicating the purposes of the law.

In practical reasoning, it is important to take into consideration what other agents will do, since this will often influence the effect of actions performed by the agent concerned. In previous treatments, the actions of others must either be assumed, or argued for using a similar form of practical reasoning. Such arguments, however, will also depend on assumptions about the beliefs, values and preferences of the other agents, and so are difficult to justify. In this paper we capture, in the form of argumentation schemes, reasoning about what others will do, which depends not on assuming particular actions, but through consideration of the expected utility (based on the promotion and demotion of values) of particular actions and alternatives. Such arguments depend only on the values and preferences of the agent concerned, and do not require assumptions about the beliefs, values and preferences of the other relevant agents. We illustrate the approach with a running example based on Prisoner's Dilemma.

Practical reasoning, reasoning about what actions should be chosen, is highly dependent both on the individual values of the agent concerned and on what others choose to do. We discuss how value based argumentation about what to do can be performed without making assumptions about the preferences of the other agents. We then show how expected utility calculations relate to the value-based argumentation approach, and express the reasoning as arguments and objections, so that they can be integrated value-based practical reasoning. We illustrate our discussion with examples of value based reasoning in public goods games as used in experimental economics and present an initial evaluation of the approach in terms of these experiments.

In this paper we present an overview of the process of argumentation with legal cases, from evidence to verdict. We identify the various different types of statement involved in the various stages, and describe how the various types relate to one another. In particular we show how we can obtain the legally accepted facts which form the basis for consideration of the law governing the cases from facts about the world. We also explain how we can determine which particular facts are relevant. In so doing we bring together several important pieces of AI and Law research and clarify their relationships.

The ANGELIC (ADF for kNowledGe Encapsulation of Legal Information from Cases) project provided a methodology for implementing a system to predict the outcome of legal cases based on a theory of the relevant domain constructed from precedent cases and other sources. The method has been evaluated in several domains, including US Trade Secrets Law. Previous systems in this domain were based on factors, which are either present or absent in a case, and favour one of the parties with the same force for every factor. Evaluations have, however, suggested that the ability to represent different degrees of presence and absence, and different strengths, could improve performance. Here we extend the methodology to allow for different degrees of presence and support, by using dimensions as a bridge between facts and factors. This new program is evaluated using a standard set of test cases.

Within characteristic function games, agents have the option of joining one of many different coalitions, based on the utility value of each candidate coalition. However, determining this utility value can be computationally complex since the number of coalitions increases exponentially with the number of agents available. Various approaches have been proposed that mediate this problem by distributing the computational load so that each agent calculates only a subset of coalition values. However, current approaches are either highly inefficient due to redundant calculations, or make the benevolence assumption (i.e. are not suitable for adversarial environments). We introduce DCG, a novel algorithm that distributes the calculations of coalition utility values across a community of agents, such that: (i) no inter-agent communication is required; (ii) the coalition value calculations are (approximately) equally partitioned into shares, one for each agent; (iii) the utility value is calculated only once for each coalition, thus redundant calculations are eliminated; (iv) there is an equal number of operations for agents with equal sized shares; and (v) an agent is only allocated those coalitions in which it is a potential member. The DCG algorithm is presented and illustrated by means of an example. We formally prove that our approach allocates all of the coalitions to the agents, and that each coalition is assigned once and only once.

In this paper we revisit reasoning with legal cases, with a view to articulating the relationships between issues, factors, facts and values, and to identifying areas for future work on these topics. We start from the different ways in which attempts have been made to go beyond a fortori reasoning from the precedent base, so that conclusions not fully justified by the precedents can be drawn. We then use a particular example domain taken from the literature to illustrate our preferred approach and to relate factors and values. From this we observe that much current work depends critically on the ascription of factors to cases in a Boolean manner, while in practice there are compelling reasons to see the presence of factors as a matter of degree. On the basis of our observations we make suggestions for the directions of future work on this topic.

Abstract Dialetical Frameworks (ADFs) are a recent development in computational argumentation which are, it has been suggested, a fruitful way of implementing theories of case law expressed in terms of factors. In this paper we evaluate this proposal, by representing the CATO analysis using ADFs. We evaluate the ease of implementation, the efficacy of the resulting program, ease of refinement of the program, transparency of the reasoning, relation to formal argumentation techniques, and transferability across domains.

In this article we offer a formal account of reasoning with legal cases in terms of argumentation schemes. These schemes, and undercutting attacks associated with them, are formalized as defeasible rules of inference within the ASPIC+ framework. We begin by modelling the style of reasoning with cases developed by Aleven and Ashley in the CATO project, which describes cases using factors, and then extend the account to accommodate the dimensions used in Rissland and Ashley's earlier HYPO project. Some additional scope for argumentation is then identified and formalized.

In this paper we describe a tool which supports the analysis of arguments in the legal domain for the purpose of building computational models that use factor-based reasoning (FBR).

In this paper we extend an argumentation scheme for practical reasoning with values based on Action-based Alternating Transition Systems. While the original scheme considers only arguments arising from the immediately next state, our proposals will enable long term considerations to be taken into account. We consider the various reasons for and against performing an action that arise from these longer term considerations, and develop a new set of argumentation schemes for practical reasoning which allows a clearer separation between facts, values and preferences, and more precise targeting of attacks.

By gaining insight into the structure and behaviours of objects drawn at random from a general class, it is often possible to develop algorithms and techniques which ameliorate the computational difficulty of decision questions arising in the general case. In this paper we present a number of approaches for the random generation of value-based argumentation frameworks (VAFs) built on n arguments and using k values. Via an empirical study we consider the behaviour of the associated random VAFs with respect to the issue of how many arguments within them have the property of being “objectively accepted”. Our studies indicate that the property of having no objectively accepted argument exhibits a so-called “phasetransition effect”, similar in nature to those observed in many other well-established AI studies.

In recent years a powerful generalisation of Dung's abstract argumentation frameworks, Abstract Dialectical Frameworks (ADF), has been developed. ADFs generalise the abstract argumentation frameworks introduced by Dung by replacing Dung's single acceptance condition (that all attackers be defeated) with acceptance conditions local to each particular node. Such local acceptance conditions allow structured argumentation to be straightforwardly incorporated. Related to ADFs are prioritised ADFs, which allow for reasons pro and con a node. In this paper we show how these structures provide an excellent framework for representing a leading approach to reasoning with legal cases.

This paper describes an on-going project investigating the use of crowdsourcing in policy consultation. We see this as particularly useful in the early consultation stages (e.g. White Paper) when the opinions of the public are sought to determine policy objectives. The project involves a number of discrete stages: thus far we have looked especially at question design, the generation of suitable test data and the suitability of various aggregation algorithms. On the basis of these results we can design software to collect opinions, and generate arguments. The test data will allow the arguments to be evaluated with respect to variously composed populations.

A Dung argumentation framework (AF) is a pair (A,R): A is a set of abstract arguments and R ⊆ A×A is a binary relation, so-called the attack relation, for capturing the conflicting arguments. Labeling based algorithms for enumerating extensions (i.e. sets of acceptable arguments) have been set out such that arguments (i.e. elements of A) are the only subject for labeling. In this paper we present implemented algorithms for listing extensions by labeling attacks (i.e. elements of R) along with arguments. Specifically, these algorithms are concerned with enumerating all extensions of an AF under a number of argumentation semantics: preferred, stable, complete, semi stable, stage, ideal and grounded. Our algorithms have impact, in particular, on enumerating extensions of AF-extended models that allow attacks on attacks. To demonstrate this impact, we instantiate our algorithms for an example of such models: namely argumentation frameworks with recursive attacks (AFRA), thereby we end up with unified algorithms that enumerate extensions of any AF/AFRA.

This paper describes a computer assisted interview for children and vulnerable adults. The system implements a “triadic interview” interaction since it is used as a focus point between the child and the clinician, whose dialogue is mediated by the tool. The tool helps children express their feelings and experiences, by making use of an “emotion palette” and a set of sub-tools developed on paper by children and transformed into computer based depictions. The tool has been extensively evaluated in clinical practice, providing a strong indication of its ability to improve the quality of the interaction with children.

Argument schemes can provide a means of explicitly describing reasoning methods in a form that lends itself to computation. The reasoning required to distinguish cases in the manner of CATO has been previously captured as a set of argument schemes. Here we present argument schemes that encapsulate another way of reasoning with cases: using preferences between social values revealed in past decisions to decide cases which have no exact matching precedents when the cases are described in terms of factors. We provide a set of schemes, with variations to capture different ways of comparing sets and varying degrees of promotion of values; we formalise these schemes; and we illustrate them with some examples.

Citizens have a variety of ways to consult with their representatives about policy proposals, seeking justifications, objecting to all or part of it, or making a counter-proposal. For the first, the representative needs only to state a justification. For the second, the representative would want to understand the objections, which may involve asking some questions. For the third, the citizen would have to provide a well formulated proposal that can then be critiqued from the standpoint of the government's own policy proposal. At the end of such a consultation, users will have aired their proposals, understood the implications, and received feedback on how their proposals contrast to that of the government.

In this paper we provide a structured analysis of US Supreme Court Oral Hearings to enable identification of the relevant issues, factors and facts that can be used to construct a test to resolve a case. Our analysis involves the production of what we term ‘argument component trees’ (ACTs) in which the issues, facts and factors, and the relationship between these, are made explicit. We show how such ACTs can be constructed by identifying the speech acts that are used by the counsel and Justices within their dialogue. We illustrate the application of our analysis by applying it to the oral hearing that took place for the case of Carney v. California, and we relate the majority and minority opinions delivered in that case to our ACTs. The aim of the work is to provide a formal framework that addresses a particular aspect of case-based reasoning: enabling the identification and representation of the components that are used to form a test to resolve a case and guide future behaviour.

In previous work we presented argumentation schemes to capture the CATO and value based theory construction approaches to reasoning with legal cases with factors. We formalised the schemes with ASPIC+, a formal representation of instantiated argumentation. In ASPIC+ the premises of a scheme may either be a factor provided in a knowledge base or established using a further argumentation scheme. Thus far we have taken the factors associated with cases to be given in the knowledge base. While this is adequate for expressing factor based reasoning, we can further investigate the justifications for the relationship between factors and facts or evidence. In this paper we examine how dimensions as used in the HYPO system can provide grounds on which to argue about which factors should apply to a case. By making this element of the reasoning explicit and subject to argument, we advance our overall account of reasoning with legal cases and make it more robust.

We introduce a derivative of Dung's seminal abstract argumentation frameworks (afs) through which distinctive features both of Dung's semantics and so-called “value-based” argumentation frameworks (vafs) may be captured. These frameworks, which we describe as uniform afs, thereby recognise that, in some circumstances, arguments may be deemed acceptable, not only as a consequence of subjective viewpoints (as are modelled by the concept of audience in vafs) but also as a consequence of “value independent” acceptance of other arguments: for example in the case of factual statements. We analyse divers acceptability conditions for arguments in uniform afs and obtain a complete picture for the computational complexity of the associated decision questions. Amongst other results it is shown that reasoning in uniform afs may pose significantly greater computational challenges than either standard or value-based questions, a number of problems being complete for the third level of the polynomial hierarchy.

In this paper we offer a proposal to enable agents to discuss the suitability of plans based on an argumentation scheme and associated critical questions. The detail encompassed by the argumentation scheme means that there is a large number of critical questions, and so dialogues may in principle be very lengthy. To improve the efficiency of dialogues we present two strategies for selecting questions. We have implemented the system and here present results showing how both strategies are effective in reducing the number of questions required to reach agreement, although their relative effectiveness is dependent on characteristics of the problem.

For Dung's theory of abstract argumentation, algorithms have been introduced for enumerating all preferred extensions. Two specific approaches have been set out that are based on labeling arguments as: IN, OUT or UNDEC. The purpose of this paper is to improve the two existing approaches by introducing two enhancements. Firstly, we employ two more informative labels. Secondly, by using these additional labels, we describe a new scheme for how the arguments' labels change in the course of computing the preferred extensions. Supported by empirical evaluation, we argue that these modifications accelerate computations. Moreover, we show how to apply the new algorithm in the context of value-based frameworks for persuasive argument, and hence, it appears that the new algorithm is usable in other formalisms extending Dung's model.

From theoretical computational perspectives, decision problems in Dung's abstract argumentation frameworks (AFs) are either polynomial solvable or intractable. To investigate practical efficiency, theoretical evaluation of applied algorithms does not necessarily reveal performance dissimilarities. Although experimental analysis of algorithms is a well-established alternative exploited in other domains, such methodology is given a little attention in the context of AFs. The main purpose of this paper is to give an example of how such experiments can be conducted to get meaningful conclusions about algorithms' behavior in situations where theoretical analysis might be of little help. To this end, we pick an extended model of AFs as a case study to empirically examine the efficiency of algorithms related to the acceptability of arguments.

In this paper we examine the problem of practical reasoning utilising Atkinson et al.'s argument scheme in situations where differences as to the desirability of social value promotion exists. We focus on the situation where a single speaker attempts to persuade a set of listeners to undertake an action, assuming that the speaker has only a probabilistic model of the listener's mental state. To achieve this, we provide a new mapping from Atkinson's argumentation scheme to a VAF, introducing a new epistemic level to the VAF. We then present a simple protocol for the interaction between a single speaker and the set of listeners. Finally, situations in which strategic reasoning is necessary for the speaker are identified.

Trust is a natural mechanism by which an autonomous party can deal with the inherent uncertainty regarding the behaviors of other parties and the uncertainty in the information it shares with those parties. Trust is thus crucial in any decentralized system. We build on recent efforts to use argumentation to reason about trust. Specifically, we provide a set of schemes, abstract patterns of reasoning that apply in multiple situations, geared toward trust. We describe, in the form of a set of critical questions, the situations in which the schemes may default.

Since the seminal work of Dung, Argumentation Frameworks have been shown to find solutions to n-person cooperative games. In multi-agent systems, decentralised methods for multi-agent system coalition structure generation have been proposed. This paper offers the first dialogue game that utilises argumentation frameworks to find a coalition structure and a payoff vector in a decentralised manner. The payoff vector found is in the core set of stable solutions if the core is non-empty. This dialogue game also puts restrictions on the payoff vectors that can be proposed so that the most unfair ones are discarded. Lastly an algorithm is described that allows the agents to find out if the core is empty.

Argumentation is key to understanding and evaluating many texts. The arguments in the texts must be identified; using current tools, this requires substantial work from human analysts. With a rule-based tool for semi-automatic text analysis support, we facilitate argument identification. The tool highlights potential argumentative sections of a text according to terms indicative of arguments (e.g. ‘suppose’ or ‘therefore’) and domain terminology (e.g. camera names and properties). The information can be used by an analyst to instantiate argumentation schemes and build arguments for and against a proposal. The resulting argumentation framework can then be passed to argument evaluation tools.

Domain models have proven useful as the basis for the construction and evaluation of arguments to support deliberation about policy proposals. Using a model provides the means to systematically examine and understand the fine-grained objections that individuals might have about the policy. While in previous approaches, a justification for a policy proposal is presented for critique by the user, here, we reuse the domain model to invert the roles of the citizen and the Government: a policy proposal is elicited from the citizen, and a software agent automatically and systematically critiques it relative to the model and the Government's point of view. Such an approach engages citizens in a critical dialogue about the policy actions, which may lead to a better understanding of the implications of their proposals and that of the Government. A web-based tool that interactively leads users through the critique is presented.

Semantic models have received little attention in recent years, much of their role having been taken over by developments in ontologies. Ontologies, however, are static, and so have only a limited role in reasoning about domains in which change matters. In this paper, we focus on the domain of policy deliberation, where policy decisions are designed to change things to realise particular social values. We explore how a particular kind of state transition system can be constructed to serve as a semantic model to support reasoning about alternative policy decisions. The policy making process includes stages that support the construction of a model, which can then be exploited in reasoning. The reasoning itself will be driven by a particular argumentation scheme for practical reasoning, and the ways in which arguments based on this scheme can be attacked and evaluated. The evaluation provides alternative policy positions. The semantics underpin a current web-based implementation, designed to solicit structured feedback on policy proposals.

In this paper we offer an account of reasoning with legal cases in terms of argumentation schemes. These schemes, and undercutting attacks associated with them, are expressed as defeasible rules of inference that will lend themselves to formalisation within the AS-PIC+ framework. We begin by modelling the style of reasoning with cases developed by Aleven and Ashley in the CATO project, which describes cases using factors, and then extend the account to accommodate the dimensions used in Rissland and Ashley's earlier HYPO project. Some additional scope for argumentation is then identified and formalised.

The paper addresses the extraction, formalisation, and presentation of public policy arguments. Arguments are extracted from documents that comment on public policy proposals. Formalising the information from the arguments enables the construction of models and systematic analysis of the arguments. In addition, the arguments are represented in a form suitable for presentation in an online consultation tool. Thus, the forms in the consultation correlate with the formalisation and can be evaluated accordingly. The stages of the process are outlined with reference to a working example.

Many websites nowadays allow social networking between their users in an explicit or implicit way. In this work, we show how the theory of argumentation schemes can provide a valuable help to formalize and structure on-line discussions and user opinions in decision support and business oriented websites that hold social networks among their users. A real study case is considered and analysed. Then, guidelines for website and system design are provided to enhance social decision support and recommendations with argumentation.

Motives play an important role at every stage of a crimi-nal investigation. In this research abstract we provide an overview of an account of motivations based on a general approach to practical reasoning. 1.

In this paper we present an approach to abductive reasoning in law by examining it in the context of an argumentation scheme for practical reasoning. We present a particular scheme, based on an established scheme for practical reasoning, that can be used to reason abductively about how an agent might have acted to reach a particular scenario, and the motivations for doing so. Plausibility here depends on a satisfactory explanation of why this particular agent followed these motivations in the particular situation. The scheme is given a formal grounding in terms of Action-based Alternating Transition Systems and we illustrate the approach with a running legal example.

In this paper we examine some previous AI and Law attempts to characterise standards of proof, and relate these to the notions of acceptability found in argumentation frameworks, an approach which forms the basis of much recent work on argumentation. We distinguish between the justification of facts and the justication of choices relating to the law and its interpretation. Standards of proof most naturally arise in connection with facts, but points of law have analogous degrees of justification.

In this paper we apply a general account of practical reasoning to arguing about legal cases. In particular, we describe how the reasoning in one very well known property law case can be reconstructed in terms of our account. We begin by summarising our general approach which uses instantiations of an argumentation scheme to provide presumptive justifications for actions, and critical questions to identify arguments which attack these justifications. These arguments and attacks are organised into argumentation frameworks to identify the status of individual arguments. Different beliefs about, and perspectives on, the issue are represented by different agents based on the Belief-Desire-Intention model, and conditions under which these agents may advance justifications and attack them are described. We model the different views of our case in these terms, describe the resulting argumentation frameworks, and relate them to the original majority and dissenting opinions. We contend that this approach both shows the worth of the general approach and its applicability to the legal domain.

In this paper we consider legal reasoning as a species of practical reasoning. As such it is important both that arguments are considered in the context of competing, attacking and supporting arguments, and that the possibility of rational disagreement is accommodated. We present two formal frameworks for considering systems of arguments: the standard framework of Dung, and an extension which relates arguments to values allowing for rational disagreement. We apply these frameworks to modelling a body of case law, explain how the frameworks can be generated to reconstruct legal reasoning in particular cases, and describe some tools to support the extraction of the value related knowledge required from a set of precedent cases.