6.1.1 Lazylav

• Name:
  Lazy lavender
• Keywords:
  Reproducibility, Multi-agent, Simulation
• Scientific Description:
  Lazy lavender aims at supporting mOeX's research on simulating knowledge evolution. It is not a general purpose simulator. However, it features some methodological innovations in term of facilitating publication, recording, and replaying of experiments.
• Functional Description:
  Lazy Lavender is a simulation environment for cultural knowledge evolution, i.e. running randomised experiments with agent adjusting their knowledge while attempting to communicate. It can generate detailed report and data from the experiments and directions to repeat them.
• Release Contributions:

  Lazy is continuously evolving and do not feature stable releases.

  Instead, use git hashes to determine which version is used in a simulation.

• News of the Year:
  In 2021, we developed multi-task agents as well as agent capability to reproduce and transmit knowledge to their offspring.
• URL:
  https://­gitlab.­inria.­fr/­moex/­lazylav/
• Publications:
  hal-03426130, hal-01661140, hal-01661139, hal-01180916
• Contact:
  Jerome Euzenat
• Participants:
  Jerome Euzenat, Yasser Bourahla, Iris Lohja, Fatme Danash, Irina Dragoste, Andrea Kalaitzakis

6.1.2 Alignment API

• Keywords:
  Ontologies, Alignment, Ontology engineering, Knowledge representation
• Scientific Description:

  The API itself is a Java description of tools for accessing the common format. It defines five main interfaces (OntologyNetwork, Alignment, Cell, Relation and Evaluator).

  We provide an implementation for this API which can be used for producing transformations, rules or bridge axioms independently from the algorithm that produced the alignment. The proposed implementation features: - a base implementation of the interfaces with all useful facilities, - a library of sample matchers, - a library of renderers (XSLT, RDF, SKOS, SWRL, OWL, C-OWL, SPARQL), - a library of evaluators (various generalisation of precision/recall, precision/recall graphs), - a flexible test generation framework that allows for generating evaluation data sets, - a library of wrappers for several ontology APIs , - a parser for the format.

  The API implementation provides an extended language for expressive alignments (EDOAL). EDOAL supports many types of restrictions inspired from description logics as well as link keys. It is fully supported for parsing and serialising in XML. It also provide other serialisers, to OWL and SPARQL queries in particular.

  To instanciate the API , it is sufficient to refine the base implementation by implementing the align() method. Doing so, the new implementation will benefit from all the services already implemented in the base implementation.

• Functional Description:
  Using ontologies is the privileged way to achieve interoperability among heterogeneous systems within the Semantic web. However, as the ontologies underlying two systems are not necessarily compatible, they may in turn need to be reconciled. Ontology reconciliation requires most of the time to find the correspondences between entities (e.g. classes, objects, properties) occurring in the ontologies. We call a set of such correspondences an alignment.
• Release Contributions:

  See release notes.

  This is the last release made from gforge svn repository. After it, the Alignment API is hosted by gitlab and versioned with git. It may well be the last formal release, clone from the repo instead.

  The Alignment API compiles in Java 11 (jars are still compiled in Java 8).

• News of the Year:
  In 2021, the painful effort of transferring the Alignment API and its documentation from gforge to gitlab has been completed. As a consequence, it is not planned anymore to issue formal releases (checking out from git and refer to the git hash should be the thing to do).
• URL:
  https://­moex.­gitlabpages.­inria.­fr/­alignapi/
• Publications:
  hal-00825931, hal-00781018
• Contact:
  Jerome Euzenat
• Participants:
  Armen Inants, Chan Le Duc, Jérôme David, Jerome Euzenat, Jérôme Pierson, Luz Maria Priego-Roche, Nicolas Guillouet

6.1.3 LinkEx

• Keywords:
  LOD - Linked open data, Data interlinking, Formal concept analysis
• Functional Description:
  LinkEx implements link key candidate extraction with our initial algorithms, formal concept analysis or pattern structures. It can extract link key expressions with inverse and composed properties and generate compound link keys. Extracted link key expressions may be evaluated using various measures, including our discriminability and coverage. It can also evaluate them according to an input link sample. The set of candidates can be rendered within the Alignment API's EDOAL language or in dot.
• URL:
  https://­gitlab.­inria.­fr/­moex/­linkex
• Publications:
  hal-02168775, hal-01179166
• Author:
  Jérôme David
• Contact:
  Jérôme David

7.1.1 Knowledge transmission across agent generations

Participants: Manuel Atencia [Correspondent], Yasser Bourahla [Correspondent], Jérôme Euzenat.

Last year, we designed a two-stage experiment in which (1) agents learn ontologies based on examples of decision making, and (2) then they interactively compare their decisions on different objects and adapt their ontologies when they disagree. We showed that agents indeed reduce interaction failure, most of the time they improve the accuracy of their knowledge about the environment, and they do not necessarily opt for the same ontology 8. In cultural evolution, this is considered as horizontal knowledge transmission between agents. Other work has shown that variation generated through vertical, or inter-generation, transmission allows agents to exceed that level 14. Such results have been obtained under the drastic selection of agents allowed to transmit their knowledge or introducing artificial noise during transmission. In order to study the impact of such measures on the quality of transmitted knowledge, we combined the settings of these two previous work and relaxed their assumptions (no strong selection of teachers, no fully correct seed, no introduction of artificial noise). Under this setting, we confirmed that vertical transmission improves on horizontal transmission even without drastic selection and oriented learning. We also showed that horizontal transmission is able to compensate for the lack of parent selection if it is maintained for long enough.

This work is part of the PhD thesis of Yasser Bourahla.

7.1.2 Pluripotent agents

Participants: Jérôme Euzenat [Correspondent], Andrea Kalaitzakis [Correspondent].

The work on cultural knowledge evolution concentrated on agents performing a single task 8. This is not a natural condition, thus we are developing agents able to carry out several tasks and to adapt their knowledge with the same protocol as before (§7.1.1). We studied the impact of having different tasks on the knowledge elaborated by agents performing such tasks. We showed that agents tackling several tasks are more successful than their single task counterparts: the higher the number of tasks, the higher the average accuracy demonstrated by a population of agents.

This work is part of the PhD thesis of Andreas Kalaitzakis.

7.1.3 Modelling cultural evolution in dynamic epistemic logic

Participants: Manuel Atencia [Correspondent], Jérôme Euzenat [Correspondent], Line van den Berg.

Ontology alignments enable agents to communicate while preserving heterogeneity in their knowledge. Alignments may not be provided as input and should be able to evolve when communication fails or when new information contradicting the alignment is acquired. The Alignment Repair Game (ARG) has been proposed for agents to simultaneously communicate and repair their alignments through adaptation operators when communication failures occur 2. ARG has been evaluated experimentally and the experiments showed that agents converge towards successful communication and improve their alignments. However, whether the adaptation operators are formally correct, complete or redundant could not be established by experiments. We introduced Dynamic Epistemic Ontology Logic (DEOL) to answer these questions. It allows us (1) to express the ontologies and alignments used via a faithful translation from ARG to DEOL, (2) to model the ARG adaptation operators as dynamic modalities and (3) to formally define and establish the correctness, partial redundancy and incompleteness of the adaptation operators in ARG 5.

This work is part of the PhD thesis of Line van den Berg defended in October 2021 12.

7.1.4 Populations

Participants: Manuel Atencia [Correspondent], Yidong Huang [Correspondent], Jérôme Euzenat [Correspondent].

The population version of the alignment repair game involves several populations of agents using the same ontologies. At given intervals, agents of the same population compare their alignments and build a consensus that they have to integrate with their own alignements. Initially it was hypothesised that such knowledge transmission can help agent achieve faster convergence, but the results suggest otherwise. Agents regularly replace their (repaired) local alignment by the (conservative) population consensus which slows down the convergence significantly. We have proposed three different ways in which agents can be critical towards the population consensus: They will adopt the consensus either based on a probability law, based on the distance between the consensus and their local alignment or based on the memory that the agent have of discarded correspondences. We tested all three approaches and found that agents can achieve faster convergence and produce alignments of comparable quality.

7.1.5 Experiment reproducibility

Participants: Yasser Bourahla [Correspondent], Clarisse Deschamps [Correspondent], Jérôme Euzenat [Correspondent].

We had maintained a registry of experiment descriptions since the beginning of the experimental cultural knowledge evolution activity. This registry was very dependent on gforge.inria.fr in order to store software (svn/git), data, experiment descriptions and analysis (wiki). The shutdown of gforge forced us to reconsider this approach. We built a new independent experimental workflow relying on git (for versioning and accountability), jupyter (for result analysis and experiment display), docker (for portable processing) and zenodo (for long-term storage of large experiment results). All previous experiments have been transferred and they are currently been repurposed as jupyter notebooks. The process and results are visible on the https://sake.re web site.

7.2.1 Strategies for identifying high-quality link keys

Participants: Jérôme David [Correspondent].

Link keys correspond to closed sets of a specific Galois connection and can be discovered thanks to a concept analysis algorithm (FCA, RCA or pattern structure). Given a pattern concept lattice whose intents are link key candidates, we aim at identifying the most relevant candidates with respect to adapted quality measures. To achieve this task, we introduced the Sandwich algorithm which is based on a combination of dual bottom-up and top-down strategies for traversing the pattern concept lattice 7. The output of the Sandwich algorithm is a partially ordered set of the most relevant link key candidates.

We also introduced the notion of non-redundant link key candidate 6. Some link keys are redundant from an extensional standpoint when the closure of the sameAs relation on the links they induce is the same as another link key.

This work is part of the PhD thesis of Nacira Abbas, co-supervised by Amedeo Napoli (LORIA).

7.2.2 Fixed-point semantics for a reduced version of relational concept analysis

Participants: Jérôme Euzenat [Correspondent].

We have used relational concept analysis (RCA) to extract link keys. This led us to notice that when there exist circular dependencies between objects, it extracts a unique stable concept lattice family grounded on the initial formal contexts. However, other stable families may exist whose structure depends on the same relational context. These may be useful in applications that need to extract a richer structure than the minimal grounded one. This year, we considered this issue in a reduced version of RCA, which only retains the relational structure. We redefined the semantics of RCA in terms of concept lattice families closed by a fixed-point operation induced by this relational structure. We showed that these families admit a least and greatest fixed point and that the well-grounded RCA semantics is characterised by the least fixed point 9. We then characterised the interesting lattices as the self-supported fixed points. We provided an algorithm to compute the greatest fixed point (dual to the RCA algorithm) and discussed strategies to extract all self-supported fixed points 13. This approach also applies to RCA in general, though we still have to provide the proofs.

7.2.3 An effective algorithm for reasoning with link keys

Participants: Manuel Atencia [Correspondent], Khadija Jradeh [Correspondent].

Link keys can be thought of as axioms in a description logic. It is thus possible to reason with link keys in combination with ontologies and alignments. In the previous years, we have designed a worst-case optimal tableau algorithm, based on compressed tableau and directed by rule application, for reasoning in the description logic $\mathrm{𝒜ℒ𝒞}$ with individuals and link keys. This algorithm is sound, complete, and has exponential complexity. A proof of concept for this algorithm has been implemented and evaluations were carried out. These evaluations show the importance of reasoning for the task of data interlinking. In particular, it is possible to use the reasoner, during link key extraction, to discard link keys introducing inconsistencies.

This work is part of the PhD thesis of Khadija Jradeh, co-supervised by Chan Le Duc (LIMICS).

8.1.1 Horizon Europe

8.2.1 Elker

Participants: Manuel Atencia [Correspondent], Jérôme David, Jérôme Euzenat, Khadija Jradeh.

mOeX coordinates the Elker project:

• Program:
  ANR-PRC
• Title:
  Extending link keys: extraction and reasoning
• Partner Institution(s):
  • Université Grenoble Alpes (coordinator)
  • INRIA Nancy Lorraine (Orpailleur)
  • Université Paris 13
• Duration:
  October 2017–September 2022
• Web site:
  https://project.inria.fr/elker/
• Abstract:
  The goal of Elker is to extend the foundations and algorithms of link keys (see §3.2) in two complementary ways: extracting link keys automatically from datasets and reasoning with link keys.

8.2.2 Miai

Participants: Manuel Atencia [Correspondent], Jérôme David [Correspondent], Jérôme Euzenat [Correspondent], Yasser Bourahla, Andreas Kalaitzakis, Line van den Berg.

mOeX holds the MIAI Knowledge communication and evolution chair

• Program:
  ANR-3IA
• Title:
  Multidisciplinary institute in artificial intelligence
• Partner Institution(s):
  • Université Grenoble Alpes (coordinator)
• Duration:
  July 2019–December 2023
• Web site:
  https://miai.univ-grenoble-alpes.fr
• Abstract:
  The MIAI Knowledge communication and evolution chair aims at understanding and developing mechanisms for seamlessly improving knowledge (see §3.3). It studies the evolution of knowledge in a society of people and AI systems by applying evolution theory to knowledge representation.

Member of organizing committees

• Jérôme Euzenat had been organiser of the 16th Ontology matching workshop of the 21th ISWC, held online, 2021 (with Pavel Shvaiko, Ernesto Jiménez Ruiz, Cássia Trojahn dos Santos and Oktie Hassanzadeh) 11
• Manuel Atencia had been member of the organizing committee of “Journée Thématique EGC & IA: Évolution et dynamique des connaissances formelles”.

9.1.1 Scientific events: selection

9.1.2 Journal

9.1.3 Invited talks

• Jérôme Euzenat participated to the “Autonomous agents on the web” Dagstuhl seminar (Online, 2021-02-14-19) 10
• Line van den Berg gave a Logic and Interactive Rationality Seminar (LIRa, Amsterdam (NL), 2021-03-25) on “Multi-Agent Knowledge Evolution in Dynamic Epistemic Logic”.
• Jérôme Euzenat gave an invited talk to the 5ème Journée AFIA-EGC Extraction et gestion des connaissances et intelligence artificielle (Grenoble (FR), 2021-05-18) on “Knowledge evolution: a computational cultural knowledge evolution perspective”.
• Line van den Berg gave a Bern Logic Seminar (Bern (CH), 2021-10-14) on “Cultural Knowledge Evolution in Dynamic Epistemic Logic”.
• Jérôme Euzenat gave an invited talk to the Ontocommons online workshop: “Ontology commons addressing challenges of the industry 5.0 transition” (Online, 2021-11-02) on “Introduction to ontology matching and alignments”.

9.1.4 Leadership within the scientific community

• Jérôme Euzenat is member of the scientific council of the cnrs GDR on Formal and Algorithmic Aspects of Artificial intelligence.
• Jérôme Euzenat is EurAI fellow.
• Jérôme David is member of the board of the Extraction and gestion des connaissances (Knowledge extraction and management) conference series.

9.1.5 Research administration

• Jérôme Euzenat is member of the COS (Scientific Orientation Committee) of INRIA Grenoble Rhône-Alpes
• Jérôme David is member of the "Commission du développement technologique" of INRIA Grenoble Rhône-Alpes
• Jérôme David is member of the LIG laboratory council

9.2.1 Teaching

9.2.2 Supervision

• Nacira Abbas, “Link key extraction and relational concept analysis”, in progress since 2018-10-01 (Jérôme David and Amedeo Napoli)
• Khadija Jradeh, “Reasoning with link keys”, in progress since 2018-10-01 (Manuel Atencia and Chan Le Duc)
• Line van den Berg, “Knowledge Evolution in Agent Populations”, defended 2021-10-29 (Manuel Atencia and Jérôme Euzenat) 12
• Yasser Bourahla, “Evolving ontologies through communication”, in progress since 2019-10-01 (Manuel Atencia and Jérôme Euzenat)
• Andreas Kalaitzakis, “Effects of collaboration and specialisation on agent knowledge evolution”, in progress since 2020-10-01 (Jérôme Euzenat)
• Alban Flandin, “The benefits of forgetting knowledge”, stopped 2021-09-30 (Jérôme Euzenat and Yves Demazeau)

9.2.3 Juries

• Jérôme Euzenat had been panel member of the habilitation defense of Antoine Zimmermann (Université Jean Monnet – Saint-Étienne).
• Jérôme Euzenat had been panel chair of the PhD defense of Muideen Lawal (Université Grenoble Alpes).
• Jérôme Euzenat had been panel chair of the PhD defense of Mathieu Viry (Université Grenoble Alpes).

9.3.1 Interventions

• Introduction by all members of mOeX of the Class? game to high-school students (tenth and eleventh graders) within the Fête de la science (Science fair), Montbonnot (FR), 2021-10-05–07.

International journals

• 4 articleM.Manuel Atencia, J.Jérôme David and J.Jérôme Euzenat. On the relation between keys and link keys for data interlinking.Semantic Web – Interoperability, Usability, Applicability1242021, 547-567
  HALDOIback to text
• 5 articleL.Line van den Berg, M.Manuel Atencia and J.Jérôme Euzenat. A logical model for the ontology alignment repair game.Autonomous Agents and Multi-Agent Systems3522021, 1-32
  HALDOIback to text

International peer-reviewed conferences

• 6 inproceedingsN.Nacira Abbas, A.Alexandre Bazin, J.Jérôme David and A.Amedeo Napoli. Non-Redundant Link Keys in RDF Data: Preliminary Steps.FCA4AI 2021 - 9th Workshop on What can FCA do for Artificial Intelligence?2972Montréal/Online, CanadaAugust 2021, 1-6
  HALback to text
• 7 inproceedingsN.Nacira Abbas, A.Alexandre Bazin, J.Jérôme David and A.Amedeo Napoli. Sandwich: An Algorithm for Discovering Relevant Link Keys in an LKPS Concept Lattice.ICFCA 2021 - 16th international conference on formal concept analysis12733LNCSStrasbourg /Virtuel, FranceSpringer Verlag2021, 243-251
  HALDOIback to text
• 8 inproceedingsY.Yasser Bourahla, M.Manuel Atencia and J.Jérôme Euzenat. Knowledge improvement and diversity under interaction-driven adaptation of learned ontologies.AAMAS 2021 - 20th ACM international conference on Autonomous Agents and Multi-Agent SystemsLondon, United Kingdom2021, 242-250
  HALback to textback to text
• 9 inproceedingsJ.Jérôme Euzenat. Fixed-point semantics for barebone relational concept analysis.Proc. 16th international conference on formal concept analysis (ICFCA)ICFCA - 16th international conference on formal concept analysisStrasbourg, FranceSpringer Verlag2021, 20-37
  HALDOIback to text
• 10 inproceedingsJ.Jérôme Euzenat. The web as a culture broth for agents and people to grow knowledge.Proc. Dagstuhl seminar on Autonomous agents on the web2021 - Dagstuhl seminar on Autonomous agents on the webWadern, GermanySchloss Dagstuhl - Leibniz-Zentrum fuer Informatik2021, 40-41
  HALback to text

Edition (books, proceedings, special issue of a journal)

• 11 proceedingsP.Pavel ShvaikoJ.Jérôme EuzenatE.Ernesto Jiménez-RuizO.Oktie HassanzadehC.Cassia TrojahnOntology Matching 2021 : Proceedings of the 16th International Workshop on Ontology Matching co-located with the 20th International Semantic Web Conference (ISWC 2021).16th ISWC workshop on ontology matching (OM 2021)3063CEUR Workshop ProceedingsonlineCEUR-WS.org2021, 1-218
  HALback to text

Doctoral dissertations and habilitation theses

• 12 thesisL.Line van den Berg. Cultural knowledge evolution in dynamic epistemic logic.Université Grenoble Alpes [2020-....]October 2021
  HALback to textback to text

Reports & preprints

• 13 reportJ.Jérôme Euzenat, M.Manuel Atencia, J.Jérôme David, A.Amedeo Napoli and J.Jérémy Vizzini. Relational concept analysis for circular link key extraction.ELKERDecember 2021, 1-57
  HALback to text