SECO

Social Evaluation of Concepts (SECO)

Funds: CRC 2017 – Free University of Bolzano.

 

Principal Investigaor: Daniele Porello

Co-Investigator: Rafael Peñaloza

The problem of integrating heterogeneous views into a collective view has been extensively studied in Welfare Economics and in particular in Social Choice Theory (SCT). In recent years, SCT provided a significant amount of information about the foundational, mathematical, and also computational aspects of preference representation and aggregation. The idea of this project is to extensively apply the rich theory of preference representation and preference aggregation provided by SCT to number of crucial problems in concept and ontology evaluation. To this end, we shall approach the problem of how to elicit and formally represent users’ preferences over definitions of concepts and over the pieces of information provided by the ontology. The project is highly interdisciplinary as it comprises disciplines such as Artificial Intelligence, for the techniques required for a formally precise and computationally aware representation of concepts, Economics and Social Choice Theory, for the theory of preference representation and preference aggregation, Philosophy, for the foundational aspects of the modelling of preferences over concepts, and Cognitive Sciences, for the assessment of the adequacy of the proposed approach to cognitive users.

The project shall focus on two research directions. Firstly, we shall approach the theoretical investigation and the precise formal modelling of preferences over concepts and ontologies. To this end, the vast amount of investigations provided by recent developments of SCT, and of Judgment Aggregation contributes in providing a precise and informative theory of revealed preferences about concepts. The second direction applies the information provided by the preferences about concepts to a number of problems in the evaluation of ontologies. Three problems that provide promising felicitous applications of this methodology are the following: the problem of integrating concepts and ontologies coming from different users, the problem of debugging ontology of possible inconsistencies, and the problem of assessing the adequacy of the ontology design with respect to the users’ views.

Although Social Choice Theory is a very well-established theory of the integration of heterogeneous views, the interplay between ontologies, knowledge representation, and aggregation of preferences is largely unexplored. This project aims to fill this gap by designing a number of applications of the methodology of SCT to ontology evaluation. The project will be closely developed in collaboration with the Institute for Logic, Language, and Computation (ILLC) of the University of Amsterdam and with the Laboratory for Applied Ontology, Institute for Cognitive Sciences and Technologies (ISTC) of the Italian National Council of Research (CNR).

 

 

Publications:

D. Porello, N. Troquard, R. Peñaloza, R. Confalonieri, P. Galliani and O. Kutz. Two Approaches to Ontology Aggregation Based on Axiom Weakening. 27th International Joint Conference on Artificial Intelligence and the 23rd European Conference on Artificial Intelligence (IJCAI-ECAI 2018).
(paper)

D. Porello, N. Troquard, R. Peñaloza, R. Confalonieri, P. Galliani and O. Kutz. Social Mechanisms for the Collective Engineering of Ontologies. Seventh International Workshop on Computational Social Choice (COMSOC-2018) Troy, NY, USA, 25-27 June 2018.
(paper)

N. Troquard, R. Confalonieri, P. Galliani, R. Peñaloza, D. Porello and O. Kutz. Repairing Ontologies via Axiom Weakening. In Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18).
(paper)

D. Porello, N. Troquard, R. Confalonieri, P. Galliani, O. Kutz, and R. Peñaloza. Repairing Socially Aggregated Ontologies Using Axiom Weakening. The 20th International Conference on Principles and Practice of Multi-Agent Systems (PRIMA 2017).
(paper)

R. Confalonieri, O. Kutz, P. Galliani, R. Peñaloza, D. Porello, M. Schorlemmer, N. Troquard. Coherence, Similarity, and Concept Generalisation. Description Logics 2017.
(paper)