Znak Politechniki Warszawskiej

Home » News »

Natural Language to make Artificial Intelligence self-explaining: the members of the NL4XAI project meet for the first time

NL4XAI’, a new European project focused on Interactive Natural Language Technology for Explainable Artificial Intelligence / CiTIUS

The aim of this four-year European project is to use natural language to generate explanations for decisions made by an AI system, which are understandable to non-expert users. The project members will meet at the 4th of September in Santiago de Compostela, Spain (to the extent this is possible under COVID-19 relate restrictions) for the first training event of the network.

According to EU legislation, humans have a right to explanations of those decisions affecting them, even if artificial intelligence (AI) based systems make such decisions; however, AI-based systems (which mainly learn automatically from data), often lack the required transparency. With the aim of contributing to avoid this, scientists work currently on the development of more transparent systems, just as they do along NL4XAI, the first European Training Network (ETN) on Natural Language (NL) and Explainable AI (XAI). The main goal of this initiative is to train eleven creative, entrepreneurial and innovative early-stage-researchers (ESRs, in general PhD students with less than 4 years of research experience) who will face the challenge of designing and implementing a new generation of self-explanatory AI systems.

NL4XAI includes a broad program of training events and opportunities, ranging from network-wide events, to courses covering technical and scientific domains and transferable skills. The ESRs will receive training in natural language generation and processing, argumentation technology and interactive interfaces for explainable AI (XAI). Additionally, they will receive training in ethical and legal questions of the field and in transdisciplinary skills. Each ESR will work on their individual research project at one of the network’s host-organizations and take part in network wide training events and meetings, as well as in secondments to other beneficiaries or partners. 

Open source framework

As a result the ESRs will be well prepared to design and build XAI models that generate interactive explanations on the basis of natural language and visual tools which are intuitively understandable by non-expert users, validated by humans in specific use cases and accessible to all European citizens.

Main outcomes are to be publicly reported and integrated into a common open source software framework for XAI. In addition, those results to be exploited commercially will be protected through licenses or patents.

First training event

One of the first milestones of the NL4XAI project will take place the 4th September 2020 along the Initial Training Meeting, as apart of the NL4XAI Network-wide Training Program (NTP), which is aimed to serve as a point of encounter for all ESRs and supervisors.

This first meeting will be co-located with the 24th European Conference on Artificial Intelligence (ECAI), Europe’s premier AI Research venue, also organized by the NL4XAI coordinator, from August 29th to September 8th, 2020. ECAI provides an opportunity for researchers to present and discuss about the best AI research, developments, applications and results. This international event will allow NL4XAI ESRs to meet the community from ECAI, facilitating feedback on ongoing thesis through the participation at the Doctoral Consortium session, and exchanging perspectives with experts.

H2020 Training Network

NL4XAI is funded by the Horizon 2020 research and innovation programme, through a Marie Skłodowska-Curie grant, in the framework of the European Union’s bet for Explainable Artificial Intelligence. The network is coordinated by the research team at the Research Centre in Intelligent Technology of the University of Santiago de Compostela (CiTIUS-USC), headed by Senén Barro. NL4XAI is a joint academic-industry research network, that brings together 18 beneficiaries and partners from six different European countries (France, Malta, Poland, Spain, The Netherlands, and UK). The partners correspond to two national research institutions (IIIA-CSIC, CNRS), ten universities (University of Aberdeen, University of Dundee, L-Universitá ta’ Malta, Delft University of Technology, Utrecht University, University of Twente, Warsaw University of Technology, Université de Lorraine, Universidade de Santiago de Compostela, Universitat Autonòma de Barcelona) and six private companies (Indra, Accenture, Orange, Wizenoze, Arria, InfoSupport).

Our tasks and challenges

Dr. Hab. Katarzyna Budzynska at the Faculty of Administration and Social Sciences at the Warsaw University of Technology is a supervisor of ESR Martijn Demollin who develops customized interactive argumentation schemes for XAI. Martijn graduated from the department of Rhetoric, Argumentation Theory and Philosophy at the University of Amsterdam where he studied argumentation schemes, language and communication.

The team of our researcher wants to define, design, develop and validate a new methodology for argumentation schemes tailored for planning and handling the discourse history in interactive conversational agents aimed at explaining AI in NL. The customization will be achieved through the creation of a profile of the user and then adaptation of argumentation to match this profile. This will involve both the analysis and the representation of the use of argumentation in natural communication in the specific genre, and the application of Budzynska’s previous pioneering work on ethotic strategies in NL, i.e., on the communicative behaviour dependent on the character of the speaker.

Main challenges are: modelling context and common ground, dealing with reference expressions, implicit knowledge and embedded testimony, empowering the conversational agent with credibility.

Theoretical contributions will be validated in two use cases: explaining decision trees to non-expert users (USC) and a real use case defined in an inter-sectoral secondment (Accenture).