Modelling cultural knowledge evolution with dynamic epistemic logics and belief revision

PhD position / Sujet de thèse

Cultural knowledge evolution considers how agents can evolve their knowledge through communicating and adapting it. Knowledge adaptation may be modelled as belief revision operators.

Cultural knowledge evolution deals with the evolution of knowledge representation in a group of agents. For that purpose, cooperating agents interact with their environment and other agents. When these agents find their behaviour inadequate, which can be detected by failing to understand others, they use operators to adapt their beliefs. This framework, originating from social sciences [Mesoudi, 2011], has been considered in the context of evolving natural languages [Steels, 2012]. We have applied it to ontology alignment repair, i.e. the improvement of incorrect alignments [Euzenat, 2017] and ontology evolution [Bourahla et al., 2021]. We have shown that it converges towards successful communication through improving the intrinsic knowledge quality.

This can be thought of as agents revising their beliefs when something tells them that they may not be correct. Belief revision operators have been designed for that purpose [Fermé and Hansson, 2018]. Multi-agent dynamic epistemic-doxastic logics (DEL for short) are dedicated to describe agent knowledge and beliefs and modelling agent actions, such as communicating, through dynamic modal operators [van Ditmarsch et al., 2007]. Belief upgrade modalities available in dynamic epistemic logics enable them to revise beliefs in the face of new information and can be thought of as belief revision [Baltag and Smets, 2006; van Benthem, 2007].

This Phd position aims at investigating deeper the relationships between cultural knowledge evolution on the one hand and dynamic epistemic logics and belief revision, on the other hand. There are, at least, two possible distinct starting points for that purpose.

First, dynamic epistemic logics have been used, in a bottom-up fashion, to model specific cultural knowledge evolution experiments and to prove the properties of adaptation operators [van den Berg, 2021]. This also raised new questions about the adequacy of dynamic epistemic logics in this context.

It would be useful to take a top-down approach to understand how far the parallel between belief upgrade and belief revision can be pursued and if this is sufficient to model more elaborate cultural knowledge evolution experiments. Indeed, although dynamic epistemic logics offer belief revision, it is hidden within the semantics of communication actions instead of being independent belief revision operators. It is thus unclear that they can really be seen as pro bono revision operators, nor that they would be sufficient in all cases. For instance, agents may perform indirectly observable actions from which other agents may induce knowledge.

Some questions that arise in relation to this problem are:

Second, modal logics, such as multi-agent dynamic epistemic logics, consider the knowledge of all agents globally, but very little the connections between agent knowledge, beside epistemic-doxastic modalities (e.g. the fact that an agent knows what another agent believes). Another way to consider belief revision in the context of cultural knowledge evolution is to consider that agents cooperatively revise common knowledge, instead of revising independently their own particular knowledge and beliefs.

Relations between agent knowledge can be expressed in the form of alignments between their knowledge. These are syntactic devices that relate terms of the theories, instead of formulas. Revision of networks of ontologies made of ontologies and alignments have already been studied [Euzenat, 2015]. It would be worth considering these in the context of dynamic epistemic logics with the alignments affecting communicative actions (queries, announcements, etc.), e.g. query answers being evaluated modulo alignments or announcements being only effective if corresponding alignments exists. In complement, the outcome of communicative actions may be belief or alignment revision. This would loosely model how knowledge and communication are dependent. Hence, it would be worth understanding how communication, enabled by alignments, influences revision and how revision influences communication.

More generally, these two research directions may contribute reconnecting multi-agent dynamic epistemic logics and distributed approaches to belief revision.

References:

[Baltag and Smets, 2006] Alexandru Baltag, Sonja Smets, Dynamic belief revision over multi-agent plausibility models, In: Proc. of 6th LOFT, pp11–24. University of Liverpool, 2006
[Bourahla et. al., 2021] Yasser Bourahla, Manuel Atencia, Jérôme Euzenat, Knowledge improvement and diversity under interaction-driven adaptation of learned ontologies, Proc. 20th AAMAS, London (UK), pp242-250, 2021 https://moex.inria.fr/files/papers/bourahla2021a.pdf
[Euzenat, 2015] Jérôme Euzenat, Revision in networks of ontologies, Artificial intelligence 228:195-216, 2015 https://exmo.inria.fr/files/publications/euzenat2015a.pdf
[Euzenat, 2017] Jérôme Euzenat, Communication-driven ontology alignment repair and expansion, Proc. 26th IJCAI, Melbourne (AU), pp185-191, 2017 https://moex.inria.fr/files/papers/euzenat2017a.pdf
[Fermé and Hansson, 2018] Edoardo Fermé, Sven Ole Hansson, Belief change: introduction and overview, Springer, Cham (CH), 2018
[Mesoudi, 2011] Alex Mesoudi, Cultural evolution: how Darwinian theory can explain human culture and synthesize the social sciences, Chicago university press, Chicago (IL US), 2011 See also: Alex Mesoudi, Andrew Whiten, Kevin Laland, Towards a unfied science of cultural evolution, Behavioral and brain sciences 29(4):329–383, 2006 http://alexmesoudi.com/s/Mesoudi_Whiten_Laland_BBS_2006.pdf
[Steels, 2012] Luc Steels (ed.), Experiments in cultural language evolution, John Benjamins, Amsterdam (NL), 2012
[van den Berg, 2021] Line van den Berg, Cultural knowledge evolution in dynamic epistemic logic, Phd thesis, Université Grenoble Alpes, 2021 https://moex.inria.fr/files/theses/thesis-vandenberg.pdf
[van Benthem, 2007] Johan van Benthem, Dynamic logic for belief revision, Journal of applied non-classical logics 17(2):129-155, 2007 https://staff.fnwi.uva.nl/j.vanbenthem/DL-BR-new.pdf
[van Ditmarsch et. al., 2007] Hans van Ditmarsch, Wiebe van der Hoek, Barteld Kooi, Dynamic epistemic logic, Springer, Synthese library 337, 2007

Links:


Qualification: Master or equivalent in computer science.

Researched skills:

Doctoral school: MSTII, Université Grenoble Alpes.

Advisor: Jérôme Euzenat (Jerome:Euzenat#inria:fr).

Group: The work will be carried out in the mOeX team common to INRIA & LIG. mOeX is dedicated to study knowledge evolution through adaptation. It gathers researchers which have taken an active part these past 15 years in the development of the semantic web and more specifically ontology matching and data interlinking.

Place of work: The position is located at INRIA Montbonnot (near Grenoble) a main computer science research lab, in a stimulating research environment.

Hiring date: October 2024.

Duration: 36 months

Deadline: as soon as possible.

Contact: For further information, contact us.

Procedure: Contact us and apply to the INRIA campaign (https://recrutement.inria.fr/public/classic/fr/offres/2024-07522). See also here.

File: Provide Vitæ, motivation letter and references. It is very good if you can provide a Master report and we will ask for your marks in Master, so if you have them, you can join them.