Counterfactual Explanations for Concepts in $\mathcal{ELH}$

Knowledge bases are widely used for information management on the web, enabling high-impact applications such as web search, question answering, and natural language processing. They also serve as the backbone for automatic decision systems, e.g. for medical diagnostics and credit scoring. As stakeholders affected by these decisions would like to understand their situation and verify fair decisions, a number of explanation approaches have been proposed using concepts in description logics. However, the learned concepts can become long and difficult to fathom for non-experts, even when verbalized. Moreover, long concepts do not immediately provide a clear path of action to change one's situation. Counterfactuals answering the question"How must feature values be changed to obtain a different classification?"have been proposed as short, human-friendly explanations for tabular data. In this paper, we transfer the notion of counterfactuals to description logics and propose the first algorithm for generating counterfactual explanations in the description logic $\mathcal{ELH}$. Counterfactual candidates are generated from concepts and the candidates with fewest feature changes are selected as counterfactuals. In case of multiple counterfactuals, we rank them according to the likeliness of their feature combinations. For evaluation, we conduct a user survey to investigate which of the generated counterfactual candidates are preferred for explanation by participants. In a second study, we explore possible use cases for counterfactual explanations.

[1]  Chin-Yew Lin,et al.  Rows from Many Sources: Enriching row completions from Wikidata with a pre-trained Language Model , 2022, WWW.

[2]  N'Dah Jean Kouagou,et al.  Neural Class Expression Synthesis , 2021, ESWC.

[3]  Varun Nandkumar Golani,et al.  EvoLearner: Learning Description Logics with Evolutionary Algorithms , 2021, WWW.

[4]  N'Dah Jean Kouagou,et al.  Learning Concept Lengths Accelerates Concept Learning in ALC , 2021, Extended Semantic Web Conference.

[5]  Axel-Cyrille Ngonga Ngomo,et al.  DRILL- Deep Reinforcement Learning for Refinement Operators in ALC , 2021, ArXiv.

[6]  Wenxing Deng,et al.  Computing Views of OWL Ontologies for the Semantic Web , 2021, WWW.

[7]  Christian Meske,et al.  Explainable Artificial Intelligence: Objectives, Stakeholders, and Future Research Opportunities , 2020, Inf. Syst. Manag..

[8]  Jose M. Alonso,et al.  Generation and Evaluation of Factual and Gounterfaetual Explanations for Decision Trees and Fuzzy Rule-based Classifiers , 2020, 2020 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE).

[9]  Bernd Bischl,et al.  Multi-Objective Counterfactual Explanations , 2020, PPSN.

[10]  Gerard de Melo,et al.  Knowledge Graphs , 2020, ACM Comput. Surv..

[11]  Ian Horrocks,et al.  Correcting Knowledge Base Assertions , 2020, WWW.

[12]  Alejandro Barredo Arrieta,et al.  Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI , 2019, Inf. Fusion.

[13]  Peter A. Flach,et al.  FACE: Feasible and Actionable Counterfactual Explanations , 2019, AIES.

[14]  Thomas Pellissier Tanon,et al.  Learning How to Correct a Knowledge Base from the Edit History , 2019, WWW.

[15]  Raphaël Troncy,et al.  Searching News Articles Using an Event Knowledge Graph Leveraged by Wikidata , 2019, WWW.

[16]  Maxat Kulmanov,et al.  EL Embeddings: Geometric construction of models for the Description Logic EL ++ , 2019, IJCAI.

[17]  Pascal Hitzler,et al.  Efficient Concept Induction for Description Logics , 2018, AAAI.

[18]  Amina Adadi,et al.  Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI) , 2018, IEEE Access.

[19]  Anne-Laure Ligozat,et al.  A Corpus for Hybrid Question Answering Systems , 2018, WWW.

[20]  Fabian M. Suchanek,et al.  Are All People Married?: Determining Obligatory Attributes in Knowledge Bases , 2018, WWW.

[21]  Franco Turini,et al.  A Survey of Methods for Explaining Black Box Models , 2018, ACM Comput. Surv..

[22]  Chris Russell,et al.  Counterfactual Explanations Without Opening the Black Box: Automated Decisions and the GDPR , 2017, ArXiv.

[23]  Jens Lehmann,et al.  DL-Learner - A framework for inductive learning on the Semantic Web , 2016, J. Web Semant..

[24]  Carsten Lutz,et al.  Schema.org as a Description Logic , 2015, Description Logics.

[25]  Giuseppe Ottaviano,et al.  Fast and Space-Efficient Entity Linking for Queries , 2015, WSDM.

[26]  Jens Lehmann,et al.  Inductive Lexical Learning of Class Expressions , 2014, EKAW.

[27]  Markus Krötzsch,et al.  Wikidata , 2014, Commun. ACM.

[28]  Jens Lehmann,et al.  DL-Learner: Learning Concepts in Description Logics , 2009, J. Mach. Learn. Res..

[29]  Jens Lehmann,et al.  DBpedia: A Nucleus for a Web of Open Data , 2007, ISWC/ASWC.

[30]  Gerhard Weikum,et al.  WWW 2007 / Track: Semantic Web Session: Ontologies ABSTRACT YAGO: A Core of Semantic Knowledge , 2022 .

[31]  Ilia Stepin,et al.  A Survey of Contrastive and Counterfactual Explanation Generation Methods for Explainable Artificial Intelligence , 2021, IEEE Access.

[32]  Jens Lehmann,et al.  SML-Bench - A benchmarking framework for structured machine learning , 2019, Semantic Web.

[33]  Ian Horrocks,et al.  An Introduction to Description Logic , 2017 .

[34]  Anni-Yasmin Turhan,et al.  A Framework for Semantic-based Similarity Measures for ELH-Concepts , 2012 .

[35]  Jens Lehmann,et al.  Concept learning in description logics using refinement operators , 2009, Machine Learning.

[36]  Sebastian Brandt,et al.  Reasoning in ELH w.r.t. General Concept Inclusion Axioms , 2004 .

[37]  Aaron Sloman,et al.  Parallel Problem Solving from Nature – PPSN XVI , 2000 .