Skip to content

Research

ICR is active in several strongly interrelated areas, notably

  • Normative multi-agent systems and deontic reasoning

  • Autonomous intelligent agents and their cognitive dynamics

  • Agreement technologies and computational social choice

  • Logic-based knowledge representation and nonmonotonic reasoning

The focus is on innovative formal techniques, whether with near-term practical applications or with inspiring visionary potential. The projects are supplemented and supported by cross-topical and preparatory research activities, as well as by diverse forms of international collaboration. In addition, we apply these research results in the following domains:

Current Projects

AuReLeE (FNR-CORE): Automated Reasoning with Legal Entities
The goal of the project AuReLeE is to provide effective and general means for the automation of normative reasoning processes based on legal knowledge bases. To this end, the project will design and implement a dedicated system that combines efficient decision procedures with a flexible approach to import and re-use existing knowledge bases for their employment as underlying contexts for the normative reasoning tasks. The results of AuReLeE hence allow to full utilization of the existing legal knowledge bases’ potential for compliance checking.

FMUAT (FNR-INTER): Formal Models for Uncertain Argumentation from Text
The topic of this project is formal models for uncertain argumentation from natural language text. Based on Dung's argumentation theory, integrating uncertainty into argumentation is gaining momentum. However, to the best of our knowledge, little attention has been paid to the modelling of uncertain argumentation in which the uncertainty of arguments is obtained mainly from text (e.g. biological papers). The aim of this project is to develop theory and algorithms to formalize and evaluate the uncertain argumentation from natural language text, such that uncertain arguments represented by natural language can be formalized and their status be properly and efficiently evaluated. The project is carried out by the cooperation between the Individual and Collective Reasoning (ICR) group at the University of Luxembourg and the group of Beishui Liao of the Center for Study of Language and Cognition (CSLC) at Zhejiang University.

PRIMAT (FNR-AFR): Probabilistic reliability management and its applications in argumentation theory and tracking objects
The PRIMAT project proposes two applications of probabilistic logic, one in the field of probabilistic argumentation, and the other in the field of probabilistic spatio-temporal reasoning. The first application follows a particular line of research in abstract argumentation, concerning the formalization of various argumentation semantics in terms of logic. Such formalizations provide a uniform definition of different semantics, lay the foundations for efficient systems, and offer the potential of direct derivation of important properties. The lack of handling levels of uncertainty and reliability in argumentation called for augmenting simple argumentation frameworks with probabilities. Unlike for standard abstract argumentation, a uniform logical formalization for probabilistic argumentation frameworks does not exist yet. The first objective of the PRIMAT project is to resolve this deficiency. The second objective is the development of a logical framework for formal reasoning about tracking moving objects. One important aspect of complex systems, such as GPS systems or security monitoring systems, is their reliability. The goal is to develop a dedicated logical language in which reliability is modelled via probability, and which also incorporates temporality. The focus will be on developing a complete axiomatic system for the logic, since having a completeness theorem is the only formal way to prove the correctness of the hardware and of the software.

SOUL (FNR-AFR) Subjective and Objective Uncertainty in Description Logics
Description Logics (DLs) are a major application-oriented research topic in Knowledge Representation and AI. They are used for modeling ontologies in many different domains (e-commerce, e-science, medicine, ...). Whereas in the past, research has focused on strict taxonomies, there are a number of areas where uncertainty has to be taken into account. The present proposal plans to investigate uncertainty in DLs on a very general level. Because detailed and reliable quantitative information is not always available, it is necessary to consider not only probabilistic knowledge, but also more qualitative uncertain information. It may be represented by defeasible rules interpreted by suitable plausibility measures (possibilistic/Spohn's ranking functions), which have been investigated in nonmonotonic reasoning, but hardly applied to DLs. Particular attention will be paid to the DL-specific separation between general conditional information (TBox), and the agent's information about specific individuals (ABox). This approach becomes more challenging when dealing with uncertainty, since the objective level, presenting general shared defeasible conditional information, may conflict with the subjective level, modeling the conditional beliefs of an agent. The intermediate expressivity of DLs is an appropriate context to investigate the interaction between both levels. The goal is to develop, analyze, and evaluate methods and implementable algorithms for attributing in a justifiable and rational way degrees of plausibility/belief to A-Box assertions about specific individuals, which amounts to complete the A-Box inductively based on defeasible/uncertain information from the T-Box.

ProCRob (FNR Proof of Concept (POC) project): Programming Cognitive Robots
Today’s service robots have emerging applications in domestic, military, health-care and entertainment domains. These applications demand ever increasing levels of intelligence and autonomy forming challenges of cognitive robotics. This is a branch of robotics that aims at studying and developing robots with reasoning capabilities needed to achieve complex goals in dynamic environments. Such robots require a processing architecture that allows them to perform high-level reasoning and deliberation about their information (i.e., beliefs and knowledge) and objectives (i.e., goals to achieve) in order to decide which actions to perform. Various agent programming languages have been proposed to support the implementation of similar cognitive architectures. However, these programming languages lack necessary supports for the management of a robot’s sensory data and the execution control of its plans. The ProCRob POC project develops a framework for rapid development and prototyping of user-friendly cognitive robots for educational purposes. The project is based on the ProCRob PhD project (2011-2015).

IELT (AFR) : Information Extraction from Legislative Texts
With the growth of the internet, laws can now be easily accessed by most citizens, but with normative production increasing at European, national and regional levels, citizens and organisations need more advanced tools to understand the law within their domain of interest. Legal informatics is a growing field of research. Legislative XML, legal ontologies and reasoning for normative systems have reached a point of maturity. However, building such resources beyond narrow applications involves a prohibitively expensive level of manual effort. Advances in natural language processing tools such as part-of-speech taggers and parsers, the growing usage of statistical algorithms for handling uncertainty and the availability of semantic resources such as WordNet and FrameNet, has resulted in robust information extraction tools. Information extraction for law is an under-researched area. Legal text, particularly legislative text, has particular features that pose significant challenges - long sentences with several clause dependencies; lists, where each item are usually not standalone sentences; and references to other articles, the content of which is not quoted within the referring article. This research investigates the transformation of legislative text into normalized sentences, and information extraction of key elements for ontologies via semantic role labelling and dependency parser trees.

NORM (AFR): Norm based deontic logic
A recent trend of Deontic logic is the move from truth-based to norm-based. Several norm-based approach to deontic logic appeared in literature: input/output logic, Hansen's logic of imperatives, and Horty's default logic. The theme of this project will further develop the theories and applications of these three approaches and unify them.

MIREL (EU-RISE, EU-Marie Curie) : Mining and Reasoning with Legal Texts
The MIREL project will create an international and inter-sectorial network to define a formal framework and to develop tools for MIning and REasoning with Legal texts, with the aim of translating these legal texts into formal representations that can be used for querying norms, compliance checking, and decision support. The development of the MIREL framework and tools will be guided by the needs of three industrial partners, and validated by industrial case studies. MIREL promotes mobility and staff exchange between SMEs to academies in order to create an inter-continental interdisciplinary consortium in Law and Artificial Intelligence areas including Natural Language Processing, Computational Ontologies, Argumentation, and Logic & Reasoning.

ProLeMAS (Marie-Curie) : Processing legal language for normative Multi-Agent Systems
The ProLeMAS project reconnects the textual representation of norms in legal documents with the logical representation of their meaning, in order to improve acceptability by legal practitioners of automatic reasoning on norms. It makes a bridge between deontic logic and natural language semantics, focusing on the modalities and the defeasible conditionals conveyed by norms. More generally, ProLeMAS develops a framework with a natural language processing pipeline able to computationally obtain explicit representations from legal text that is effective and acceptable to lawyers. ProLeMAS opens a new research trend in normative Multi-Agent systems, along three dimensions. First, ProLeMAS enhances the expressive power of deontic logic to formalize the meaning of the phrases constituting sentences, including a wide range of fine-grained intra-sentence linguistic phenomena. Natural language semantics is not part of the NorMAS roadmap, although it has been identified as a critical issue by the current scientific community in deontic logic and normative systems, as witnessed by the special focus on “deontic modalities in natural language” at the DEON 2014 conference. Secondly, ProLeMAS defines a first-order decision theory able to make inferences on norms from legislation as well as agents’ goals and attitudes. Third, ProLeMAS will develop a prototype able to extract obligations from laws and codify them in the chosen object logic. No system developed so far by members of the NorMAS community is capable of processing legal documents available on the Web. The prototype that will be implemented in ProLeMAS will use and extend two specific tools: the TULE parser and the Tacitus system.

Last-JD Projects

The objective of the Joint International Doctoral (Ph.D.) Degree in Law, Science and Technology (Last-JD) is to create an interdisciplinary integrated programme in order to face the new challenges that the information society and the newly emerging technologies will increasingly pose in the future in the legal domain and the socioethical field. The Doctorate offers an innovative and up-to-date integrated programme that empowers candidates to carry out cutting-edge research in different subjects such as bioethics, law and computer science for topics that require a genuine interdisciplinary approach. It includes a period of internship in the third year, where the candidate can apply his/her studies and complement his/her research with a professional experience. Upon completion of the programme, graduates will become highly skilled researchers and professionals, matching the requirements of public and private sectors. The following projects are currently active at ICR:

  • Cloud Computing Security Issues and Its Regulation
  • The Place of Legal Ontologies in Co-regulatory Compliance
  • Transnational Interactions among Legal Systems and Defeasible Logics
  • Legal Knowledge Framework in ODR – Definition of Legal “Relevant Information” in Consumer Law Disputes: Telecommunications and Air Transport Passengers
  • Privacy Protection Model for Online Social Networks

Finished Projects

  • SIEP (FNR-INTER): Specification logics and Inference tools for verification and Enforcement of Policies

  • RATARCH (FNR-CORE): Rationalization of Architecture Related Design Decisions

  • RAT (FNR-MOBILITY): Reasoning about Agreement Technologies - 6 month sabbatical of Prof. van der Torre at CSLI Stanford

  • SGAMES (FNR-CORE): Security Games

  • MARCO (FNR-CORE) : Managing Regulatory Compliance: a Business-Centred Approach

  • LAAMI (FNR-CORE) : Logical Analysis of Market Irrationality

  • DYNARG (FNR-INTER/CNRS) : Dynamics of Argumentation

  • LINMAS (UL-POSTDOC) : Logics Integrated for Normative Multi-Agent Systems

  • TRUSTGAMES (AFR-POSTDOC) : Trust Games

  • CFAEMM (AFR-POSTDOC) : A Computational Framework for Apprehending Evolving Malware and Malware Engineers

  • CDL (ERCIM-POSTDOC) : Argument-based Contextual Defeasible Reasoning

  • ICR (UL-PHD) : Individual and Collective Reasoning

  • AASTM (UL-PHD) : Argumentation Techniques for Trust Management

  • LOSEC (AFR-PHD) : Logics for Security

Collaboration