Study with knowledge diversity measures

Short topic / Sujet de TER

We previously showed that agents, evolving their knowledge for agreeing, may preserve their diversity in the form of non equivalent ontologies. We implemented a better justified entropy-based measure that takes into account the similarity between categories of representations. We aim at reconsidering the former results under the light of the new measure.

The diversity of many phenomena is worth measuring because diversity is often related to fairness and sometimes to resilience. It is also good to measure how our actions increase or decrease diversity. We are interested in knowledge diversity: the fact that different people may held different knowledge and beliefs [1].

Measuring diversity amounts to associate a relative quantity to the knowledge diversity of a population of agents. We only consider agent populations of the same size whose individuals are distributed in different categories. Diversity may be considered from three different dimensions:

Variety: how many categories of individuals are represented? (insects, molluscs, vertebrate vs. insects, molluscs)
Balance: how many representatives of each category are there? (1 vertebrates, 99 insects vs. 50 vertebrate, 50 insects)
Disparity: how different are these categories? (33 ants, 33 bugs, 33 grasshoppers vs. 33 rabbits, 33 ants, 33 snakes)

We defined an entropy-based measure inspired from the work of Tom Leinster [2] which is parametrised by a similarity measure between the categories. It puts the emphasis of the equal dispersion of observations in these types weighted by the similarity between such types. In our case, the categories are ontologies and the similarity measure will account for the semantic similarity between them. It is parametrised by another parameter q, ranging in [0 +∞], which puts more or less emphasis on variety and balance. One population can be said to be more diverse than another if the measure of the former is higher than that of the latter for all values of q. We implemented this measure in Python [4] and it provides the expected results.

In 2021, we used a distance measure as a proxy to measure diversity. This allowed us to determine that, though the distance between agent knowledge decreased, it was not brought to zero. Thus, some knowledge diversity was preserved [3].

The goal of this work is to reevaluate this claim with the entropy-based knowledge diversity measure. It could be interesting to understand if it changes this claim, in which proportions and how.

For that purpose, it will be necessary to rerun the previous experiments as diversity cannot be computed from the collected data. Safeguarding agent ontologies at the beginning and end of the experiments will be necessary. The similarity measure can be easily defined as in function of the previous semantic distance. However, in order to compute the measure it will be necessary to determine the categories of interest: these could be the ontologies that at least one agent has, in a single experiment or in all of them, or the possible ontologies over the set of attributes.

This should require some knowledge on Java and Python programming.

The work could be expected to unfold as follows:

Studying the framework proposed in [3].
Modify the code to provide the necessary data (agent ontologies).
Compute different diversity measures on the results and analyse them.

References:

[1] Yasser Bourahla, Jérôme David, Jérôme Euzenat, Meryem Naciri, Measuring and controlling knowledge diversity, Proc. 1st JOWO workshop on formal models of knowledge diversity (FMKD), Jönköping (SE), 2022 https://moex.inria.fr/files/papers/bourahla2022c.pdf
[2] Tom Leinster, Entropy and diversity: the axiomatic approach, Cambridge university press, Cambridge (UK), 2021 https://arxiv.org/pdf/2012.02113.pdf
[3] 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

Links:
[4] Knowledge diversity notebook: https://moex.inria.fr/software/kdiv/index.html
[5] mOeX web site: https://moex.inria.fr

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

Laboratory: LIG.

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

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