MOSIG Master 2ND YEAR Research
YEAR 2018/2019

Replicator/interactor model for cultural knowledge evolution

Master topic / Sujet de master recherche

Evolution theory may be applied to knowledge which is variable, communicated and selected among populations. Generalised theories of evolution introduce the notion of replicators (generalising genes) which generate an interactor (generalising phenotype) which is subject to natural selection. Because knowledge is the cause of individual behaviour which is subject to selection, either by the individual or its environments, it is tempting to apply this model to knowledge evolution.

Cultural evolution is the application of evolution theory to culture [Messoudi 2006]. It may be addressed through multi-agent simulations [Steels 2012]. Experimental cultural evolution provides a population of agents with interaction games that are played randomly. In reaction to the outcome of such games, agents adapt their knowledge. It is possible to test hypotheses by precisely crafting the rules used by agents in games and observing the consequences.

This has been applied to the evolution of the way agents represent knowledge [Euzenat 2014; Anslow 2015; Chocron 2016]. We have applied it to ontology alignment repair, i.e., the improvement of incorrect alignments [Euzenat 2014; 2017a]. We showed that cultural repair is able to converge towards successful communication and improves the objective correctness of alignments.

Modern theories of evolution may be based on the concept of replicator/interactor which generalises that of genotype/phenotype [Dawkins 1976; Hull 1980; Wilkins 2014]. Like genes, replicators may be transmitted to the offspring almost perfectly, modulo mutation (variation). They are also the pattern for the whole organism: they determine the traits (phenotype) of individuals. Among those traits expressed by genes, some are under the pressure of the environment (interactors). Hence, depending on these traits, organisms will be more or less fit to the environment (in good shape to eat, survive, mate, have offspring). Thus selection operates indirectly on the capacity of replicators to been inherited.

The replicator/interactor model seems to fit perfectly knowledge evolution. Indeed, we transmit our knowledge (replicator) and it influences our behaviour (interactor). In turns, this behaviour is what will be selected by the environment, by others and ultimately by ourselves. The figure below explain how agents may be built following this model.

There are however differences between the genuine generalisation of genetics and cultural knowledge evolution:

The goal of this master topic is to apply explicitly the replicator/interactor model to knowledge evolution. This involves further designing agents implementing this model and experiments that highlight the properties of such agents, in particular, with respect to some of the differences noted above.

Expected contributions are firstly the design of agents applying the replictor/interactor model to knowledge as well as the study of the behaviour of such agents, e.g., their capability to evolve knowledge.

This work is part of an ambitious program towards what we call cultural knowledge evolution and may prepare to a PhD.

References:

[Anslow 2015] Michael Anslow, Michael Rovatsos, Aligning experientially grounded ontologies using language games, Proc. 4th international workshop on graph structure for knowledge representation, Buenos Aires (AR), pp15-31, 2015 [DOI:10.1007/978-3-319-28702-7_2]
[Chocron 2016] Paula Chocron, Marco Schorlemmer, Attuning ontology alignments to semantically heterogeneous multi-agent interactions, Proc. 22nd European Conference on Artificial Intelligence, Der Haague (NL), pp871-879, 2016 [DOI:10.3233/978-1-61499-672-9-871]
[Dawkins 1976] Richard Dawkins, The selfish gene, Oxford University Press, Oxford (UK), 1976 (exp. 1989)
[Euzenat 2014] Jérôme Euzenat, First experiments in cultural alignment repair (extended version), in: Proc. 3rd ESWC workshop on Debugging ontologies and ontology mappings (WoDOOM), Hersounisos (GR), LNCS 8798:115-130, 2014 https://exmo.inria.fr/files/publications/euzenat2014c.pdf
[Euzenat 2017a] Jérôme Euzenat, Communication-driven ontology alignment repair and expansion, in: Proc. 26th International joint conference on artificial intelligence (IJCAI), Melbourne (AU), pp185-191, 2017 https://moex.inria.fr/files/papers/euzenat2017a.pdf
[Hull 1980] David Hull, Individuality and selection, Annual review of ecology and systematics 11:311-332, 1980
[Mesoudi 2006] 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
[Wilkins 2014] John Wilkins, David Hull, Replication and Reproduction, The Stanford Encyclopedia of Philosophy (Spring 2014 Edition), https://plato.stanford.edu/entries/replication/

Links:


Reference number: Proposal n°2458

Master profile: M2R MOSIG, Artificial intelligence and the web profile.

Advisor: Jérôme Euzenat (Jerome:Euzenat#inria:fr) and Manuel Atencia (Manuel:Atencia#inria:fr).

Team: The work will be carried out in the mOeX team common to INRIA & Université Grenoble Alpes. mOeX is dedicated to study knowledge evolution through adaptation. It gather permanent researchers from the Exmo team which has taken an active part these past 15 years in the development of the semantic web and more specifically ontology matching.

Laboratory: LIG.

Procedure: Contact us and provide vitæ and possibly motivation letter and references.