Stefan Schlobach

AIJ Journal 2022 Journal Article

Knowledge graphs as tools for explainable machine learning: A survey

• Ilaria Tiddi
• Stefan Schlobach

This paper provides an extensive overview of the use of knowledge graphs in the context of Explainable Machine Learning. As of late, explainable AI has become a very active field of research by addressing the limitations of the latest machine learning solutions that often provide highly accurate, but hardly scrutable and interpretable decisions. An increasing interest has also been shown in the integration of Knowledge Representation techniques in Machine Learning applications, mostly motivated by the complementary strengths and weaknesses that could lead to a new generation of hybrid intelligent systems. Following this idea, we hypothesise that knowledge graphs, which naturally provide domain background knowledge in a machine-readable format, could be integrated in Explainable Machine Learning approaches to help them provide more meaningful, insightful and trustworthy explanations. Using a systematic literature review methodology we designed an analytical framework to explore the current landscape of Explainable Machine Learning. We focus particularly on the integration with structured knowledge at large scale, and use our framework to analyse a variety of Machine Learning domains, identifying the main characteristics of such knowledge-based, explainable systems from different perspectives. We then summarise the strengths of such hybrid systems, such as improved understandability, reactivity, and accuracy, as well as their limitations, e. g. in handling noise or extracting knowledge efficiently. We conclude by discussing a list of open challenges left for future research.

KR Conference 2020 Conference Paper

On Sufficient and Necessary Conditions in Bounded CTL: A Forgetting Approach

• Renyan Feng
• Erman Acar
• Stefan Schlobach
• Yisong Wang
• Wanwei Liu

Computation Tree Logic (CTL) is one of the central formalisms in formal verification. As a specification language, it is used to express a property that the system at hand is expected to satisfy. From both the verification and the system design points of view, some information content of such property might become irrelevant for the system due to various reasons, e. g. , it might become obsolete by time, or perhaps infeasible due to practical difficulties. Then, the problem arises on how to subtract such piece of information without altering the relevant system behaviour or violating the existing specifications over a given signature. Moreover, in such a scenario, two crucial notions are informative: the strongest necessary condition (SNC) and the weakest sufficient condition (WSC) of a given property. To address such a scenario in a principled way, we introduce a forgetting-based approach in CTL and show that it can be used to compute SNC and WSC of a property under a given model and over a given signature. We study its theoretical properties and also show that our notion of forgetting satisfies existing essential postulates of knowledge forgetting. Furthermore, we analyse the computational complexity of some basic reasoning tasks for the fragment CTLAF in particular.

KR Conference 2014 Short Paper

A Pragmatic Semantics for Identity in Linked Data

• Wouter Beek
• Stefan Schlobach
• Frank Van Harmelen

IS Journal 2009 Journal Article

Evaluating Thesaurus Alignments for Semantic Interoperability in the Library Domain

• Antoine Isaac
• Shenghui Wang
• Claus Zinn
• Henk Matthezing
• Lourens van der Meij
• Stefan Schlobach

Thesaurus alignments play an important role in realizing efficient access to heterogeneous cultural-heritage data. Current technology, however, provides only limited value for such access because it fails to bridge the gap between theoretical study and practical application requirements. This article explores common real-world library problems and identifies solutions that focus on the application-embedded study, development, and evaluation of matching technology.

IJCAI Conference 2007 Conference Paper

• Stefan Schlobach
• Michel Klein
• Linda Peelen

We extend traditional Description Logics (DL) with a simple mechanism to handle approximate concept definitions in a qualitative way. Often, for example in medical applications, concepts are not definable in a crisp way but can fairly exhaustively be constrained through a particular sub- and a particular super-concept. We introduce such lower and upper approximations based on rough-set semantics, and show that reasoning in these languages can be reduced to standard DL satisfiability. This allows us to apply Rough Description Logics in a study of medical trials about sepsis patients, which is a typical application for precise modeling of vague knowledge. The study shows that Rough DL-based reasoning can be done in a realistic use case and that modeling vague knowledge helps to answer important questions in the design of clinical trials.

AAAI Conference 2005 Conference Paper

Diagnosing Terminologies

• Stefan Schlobach

We present a framework for the debugging of logically contradicting terminologies, which is based on traditional modelbased diagnosis. To study the feasibility of this highly general approach we prototypically implemented the hitting set algorithm presented in (Reiter 1987), and applied it in three different scenarios. First, we use a Description Logic reasoning system as a black-box to determine (necessarily maximal) conflict sets. Then we use our own non-optimized DL reasoning engine to produce small, and a specialized algorithm to determine minimal conflict sets. In a number of experiments we show that the first method already fails for relatively small terminologies. However, based on small, or minimal conflict sets, we can often calculate diagnoses in reasonable time.

JELIA Conference 2004 Conference Paper

Explaining Subsumption by Optimal Interpolation

• Stefan Schlobach

Abstract We describe ongoing research to support the construction of terminologies with Description Logics. For the explanation of subsumption we search for particular concepts because of their syntactic and semantic properties. More precisely, the set of explanations for a subsumption \(P\sqsubseteq N\) is the set of optimal interpolants for P and N. We provide definitions for optimal interpolation and an algorithm based on Boolean minimisation of concept-names in a tableau proof for \(\mathcal{ALC}\) -satisfiability. Finally, we describe our implementation and some experiments to assess the computational scalability of our proposal.

ECAI Conference 2004 Conference Paper

Type Checking in Open-Domain Question Answering

• Stefan Schlobach
• Marius Olsthoorn
• Maarten de Rijke

IJCAI Conference 2003 Conference Paper

Non-Standard Reasoning Services for the Debugging of Description Logic Terminologies

• Stefan Schlobach
• Ronald Cornet

Current Description Logic reasoning systems provide only limited support for debugging logically erroneous knowledge bases. In this paper we propose new non-standard reasoning services which we designed and implemented to pinpoint logical contradictions when developing the medical terminology DICE. We provide complete algorithms for unfoldable ACC-TBoxes based on minimisation of axioms using Boolean methods for minimal unsatisfiability-presening sub-TBoxes, and an incomplete bottom-up method for generalised incoherence-preserving terminologies.