ONTOarg: A decision support framework for ontology integration based on argumentation
Highlights
► We show a method for performing local-as-view ontology integration in the presence of possible inconsistencies. ► The method is implemented in Defeasible Logic Programming. ► The method is apt for being implemented in the OWL language, which is the standard for Semantic Web applications. ► Some properties of the approach are presented. ► A framework is presented that could be used to implement an expert system, including a case study to show how the proposal works.
Introduction
The Semantic Web (Berners-Lee, Hendler, & Lassila, 2001) (SW) is a vision of the Web where agents can reason about resources whose meaning is assigned in terms of ontologies (Gruber, 1993). Within the Semantic Web, the OWL language has become the current standard for defining ontologies and its underlying semantics is based on Description Logics (DL) (Baader, Calvanese, McGuinness, Nardi, & Patel-Schneider, 2003), for which specialized reasoners exist (such as Racer (Haarslev & Möller, 2001) and Pellet (Parsia & Sirin, 2004)). In this context, Description Logic Programming (DLP) provides a practical approach to reason with DL ontologies, translating them into the language of logic programming (LP) (Grosof, Horrocks, Volz, & Decker, 2003). Although DLP offers several advantages in terms of efficiency and reuse of existing LP tools (such as Prolog environments), DLP is incapable of reasoning in the presence of inconsistent ontologies. Dealing with inconsistencies and potentially contradictory information is a common issue in semantic web reasoning. The semantic integration of potentially inconsistent information sources in the SW is complicated, as the knowledge engineer usually has no authority to correct foreign ontologies. Inconsistencies can also arise whenever the domains modeled are inherently inconsistent. Additionally, in many situations resources and their data are modeled in terms of ontologies whose terms can differ, so that the ontologies must be aligned to put their terms into mutual agreement (Klein, 2001).
Argumentation provides a sophisticated mechanism for the formalization of commonsense reasoning, which has found application and proven its importance in different areas of Artificial Intelligence (AI) such as legal systems, multi-agent systems, and decision support systems among others (see e.g. Bench-Capon and Dunne, 2007, Janjua and Hussain, 2012, Modgil et al., 2013, Rahwan and Simari, 2009). Intuitively, an argument can be thought of as a coherent set of statements that supports a claim. The ultimate acceptance of an argument will depend on a dialectical analysis of arguments in favor and against the claim (Rahwan & Simari, 2009). In the last decade, several frameworks for formalizing argumentative reasoning have been developed (also called generically Argumentation Systems), providing different knowledge representation and inference capabilities.
Recent research has led to the use of defeasible argumentation to model different DL reasoning capabilities when handling inconsistent ontologies, resulting in so-called δ-ontologies (Gómez, Chesñevar, & Simari, 2010). These special ontologies are based on Defeasible Logic Programming (DeLP) (García & Simari, 2004), a defeasible argumentation framework based on logic programming. In contrast with other approaches, δ-ontologies have the flexibility of allowing to assess defeasibly the membership of an individual to a concept description in the presence of a potential inconsistent ontology. However, this approach was intended for a single ontology, and fell short when dealing with different ontologies which have to be integrated in a single one. There are two kinds of ontology integration approaches, viz. global-as-view (GAV) and local-as-view (LAV) (Calvanese, Giacomo, & Lenzerini, 2001). When performing LAV integration, concepts of the local ontologies are mapped to queries over a global ontology.
In this article, we present ONTOarg, a decision support framework for modeling LAV ontology integration when the involved ontologies can be potentially inconsistent. The ontologies are expressed in the language of DL but their semantics is expressed in terms of DeLP. The alignments between the local and global ontologies are expressed as DL inclusion axioms that are also interpreted as DeLP sentences. As the ontologies are potentially inconsistent, a dialectical analysis is performed on the interpretation of both the ontologies and the mappings from the local to the global ontology. ONTOarg can be seen as a front-end expert system for the ontology engineers, which solves automatically the problem of alignments using an argumentative proof procedure.
The rest of this paper is structured as follows. In Section 2 we present the fundamentals of Description Logics and Defeasible Logic Programming. Section 3 recalls the δ-ontologies framework for reasoning with possibly inconsistent ontologies. In Section 4, we extend δ-ontologies for performing local-as-view integration of possibly inconsistent ontologies. Section 5 discusses some properties of the proposed approach. In Section 6 we compare our approach with related work. Finally Section 7 concludes the paper.
Section snippets
Description logics
Description Logics (DL) (Baader et al., 2003) are a family of knowledge representation formalisms based on the notions of concepts(unary predicates, classes) and roles (binary relations) that allow to build complex concepts and roles from atomic ones.
Let C, D stand for concepts, R for a role and a, b for individuals. Concept descriptions are built from concept names using the constructors conjunction (C ⊓ D), disjunction (C ⊔ D), complement (¬C), existential restriction (∃R.C), and value
Reasoning with inconsistent ontologies: δ-ontologies
In the presence of inconsistent ontologies, traditional DL reasoners (such as Racer (Haarslev & Möller, 2001)) issue an error message and stop further processing. Thus the burden of repairing the ontology (i.e., making it consistent) is on the knowledge engineer. However, the knowledge engineer is not always available and in some cases, such as when dealing with imported ontologies, he has neither the authority nor the expertise to correct the source of inconsistency. Therefore, we are
The ONTOarg framework: architecture
The notion of δ-ontology is powerful enough to reason defeasibly within a specific ontology with inconsistencies. However, a common problem for a knowledge engineer is associated with dealing with different δ-ontologies, which have to be unified or integrated in order to get a global understanding of the knowledge available in such ontologies, resulting in turn in the characterization of a global ontology. In the literature (Calvanese et al., 2001), the term ontology integration is associated
Some properties of the approach
We now introduce some properties of the proposed approach to ontology integration presented above. Notice that the validity of the properties is strongly related to DeLP reasoning dynamics. Property 1 Let be an ontology integration system. It cannot be the case that an individual a is a justified member of concepts C and ¬C simultaneously. Proof Suppose that both a is a justified member of both C and ¬C. Then it must be the case that there exist two warranted arguments and . But this
Related work
Next we will review some recent research in reasoning with inconsistencies in ontologies, reasoning with ontologies in logic programming and recent advances in ontology integration, contrasting existing results with our approach.
Conclusions
We have presented an approach for performing local-as-view integration of Description Logic ontologies when these ontologies can be potentially inconsistent. We have adapted the notion of ontology integration system of Calvanese et al. (2001) for making it suitable for the δ-ontology framework, presenting both formal definitions and a case study.
For reasoning with an inconsistent ontology, it can be repaired manually by the knowledge engineer or automatically (e.g., using Belief Revision (
Acknowledgements
The research was funded by LACCIR Project 1211LAC004, by PIP CONICET Projects 112-200801-02798 and 112-200901-00863, and by Sec. Gral. de Ciencia y Tecnología, Universidad Nacional del Sur.
References (61)
- et al.
Defeasible logic versus logic programming without negation as failure
Journal of Logic Programming
(2000) - et al.
Argumentation in artificial intelligence
Artificial Intelligence
(2007) - et al.
Argument-based critics and recommenders: A qualitative perspective on user support systems
Data & Knowledge Engineering (DKE)
(2006) A translation approach to portable ontologies
Knowledge Acquisition
(1993)- et al.
Web@idss argumentation-enabled web-based idss for reasoning over incomplete and conflicting information
Knowledge-Based Systems
(2012) Defeasible reasoning
Cognitive Science
(1987)- et al.
A mathematical treatment of defeasible reasoning and its implementation
Artificial Intelligence
(1992) - et al.
Frame-based argumentation for group decision task generation and identification
Decision Support Systems
(2005) - et al.
DR-Prolog: A system for defeasible reasoning with rules and ontologies on the semantic web
IEEE Transactions on Knowledge and Data Engineering
(2007) - et al.
Normal forms for defeasible logic
Argumentation-supported information distribution in a multiagent system for knowledge management
Non monotonic reasoning
On the issue of contraposition of defeasible rules
Argument-based applications to knowledge engineering
The Knowledge Engineering Review
Knowledge distribution in large organizations using defeasible logic programming
Logical models of argument
ACM Computing Surveys
Logic programming without negation as failure
Defeasible logic programming an argumentative approach
Theory and Practice of Logic Programming
Defeasible reasoning in web forms through argumentation
International Journal of Information Technology & Decision Making
Cited by (28)
A combined multi-agent system for distributed multi-project scheduling problems
2021, Applied Soft ComputingAn approach for distributed discussion and collaborative knowledge sharing: Theoretical and empirical analysis
2019, Expert Systems with ApplicationsCitation Excerpt :This latter point is of particular interest since, as mentioned in Tintarev and Masthoff (2007), qualitative explanations are relevant to user support systems, because users prefer recommendations when they can understand the reasons behind its selection. Such advantages have motivated the development of several argumentation based recommender and decision support systems: (Amgoud & Prade, 2009; Bedi & Vashisth, 2014; Gómez, Chesñevar, & Simari, 2013; Teze, Gottifredi, García, & Simari, 2015), among others. In our approach, the collaborative reasoning will be performed as an iterative process in which agents pose or reject arguments until they reach an agreement about the piece information that is being discussed about.
Explanatory dialogues with argumentative faculties over inconsistent knowledge bases
2017, Expert Systems with ApplicationsImproving remote sensing crop classification by argumentation-based conflict resolution in ensemble learning
2016, Expert Systems with ApplicationsCitation Excerpt :As technical instrumentation, we rely on Defeasible Logic Programming (DeLP) formalized by García and Simari (2004) to perform argument based reasoning. DeLP has proved to be an effective knowledge representation and dialectical-based argumentation system for real-world applications: recommender systems by Bedi and Vashisth (2014); Briguez et al. (2014), ontology reasoning by Gómez, Chesñevar, and Simari (2013), safety assurance for unmanned aerial vehicles by Gómez, Goron, Groza, and Letia (2016), relational databases by Deagustini et al. (2013), or planning by Chow, Siu, Chan, and Chan (2013). From the knowledge perspective, ensemble learning method proposed by Xu, Yao, and Li (2015) integrates only the classification results of single classifiers.
A structured argumentation system with backing and undercutting
2016, Engineering Applications of Artificial IntelligenceCitation Excerpt :The principles stated in their work are exemplified in a robotic domain where a Khepera 2 robot collects boxes scattered through the environment and therefore, should make decisions about which box must be transported next. Another concrete application of DeLP can be found in Gómez et al. (2013), where the authors propose ONTOarg, a decision support framework for performing local-as-view integration of possibly inconsistent and incomplete ontologies in terms of DeLP. Finally, in Deagustini et al. (2013) the authors deal with the problem of massively built arguments from premises obtained from relational databases and compute warrant.
Assuring safety in air traffic control systems with argumentation and model checking
2016, Expert Systems with ApplicationsCitation Excerpt :DeLP (García & Simari, 2004) is one of such formalisms, combining results from defeasible argumentation theory (Simari & Loui, 1992) and logic programming (Lloyd, 1987). DeLP is a suitable framework for building real-world applications which has proven to be particularly attractive in that context, such as clustering (Gómez & Chesñevar, 2004), intelligent web search (Chesñevar, Maguitman, & Simari, 2006; Chesñevar & Maguitman, 2004b), knowledge management (Chesñevar, Brena, & Aguirre, 2005a, 2005b), multiagent systems (Brena, Chesñevar, & Aguirre., 2006), natural language processing (Chesñevar & Maguitman, 2004a), intelligent web forms (Gómez, Chesñevar, & Simari, 2008), ontology reasoning (Gómez, Chesñevar, & Simari, 2010; Gómez et al., 2013), relational databases (Deagustini et al., 2013), planning (Chow, Siu, Chan, & Chan, 2013), recommender systems (Bedi & Vashisth, 2014; Briguez et al., 2014), machine learning (Fomina, Morosin, & Vagin, 2014), among others. At time 0, for flight f701 we only have its identification as there is only a fact flight(f701) to consider.