Franz Weitl

Towards the automated verification of semi-structured documents

The semantic consistency of context-dependent documents is an important problem in the fields of document databases and web information systems. While techniques for adapting documents to different context of use are well investigated, the consistency maintenance of context-dependent documents remains a challenge. Existing methods are specialized on specific document formats or domains, or are cumbersome and costly to apply. In this paper, we present a new approach focusing on user support by a pattern-based specification methodology. We combine temporal logic, ontologies, and a pattern-based specification approach for reaching high expressiveness, flexible applicability, and yet a high degree of usability.

Logic-based verification of technical documentation

Checking the content coherence of digital documents is the purpose of the Verdikt system which can be applied to different domains and document types including technical documentation, e-learning documents, and web pages. An expressive temporal description logic allows for the specification of content consistency criteria along document paths. Whether the document conforms to the specification can then be verified by applying a model checker. In case of specification violations, the model checker provides counterexamples, locating errors in the document precisely. Based on a sample technical documentation in the form of a web document, the general verification process and its effectiveness, efficiency, and usability are demonstrated.

Structured Counterexamples for the Temporal Description Logic ALCCTL

A new algorithm for generating counterexamples for the temporal description logic ALCCTL is presented. ALCCTL is a decidable combination of the description logic ALC and computation tree logic CTL. It extends CTL by first order quantified expressions over unary and binary predicates. Predicates and quantified expressions are required for representing properties in application domains such as structured web documents and they are frequently used in software and hardware specifications which are verified by model checking. In the case of a specification violation, existing algorithms generate counterexamples that tend to be complex yet imprecise if specifications range over sets of objects. The presented algorithm is the first algorithm for generating counterexamples for a temporal description logic that considers first order predicates and quantification. The algorithm is sound and semi-complete for ALCCTL. The generated counterexamples are both more precise and comprehensible than counterexamples generated by the previous algorithms.

Checking semantic integrity constraints on integrated Web documents

A conceptual framework for the specification and verification of constraints on the content and narrative structure of documents is proposed. As a specification formalism we define CTL \({_\mathcal {DL}}\) which is a new version of the temporal logic CTL extended with description logic concepts. In contrast to existing solutions our approach allows for the integration of ontologies to achieve interoperability and abstraction from implementation aspects of documents. This makes it specifically suitable for the integration of heterogenous and distributed information resources in the semantic web.

Verifying the consistency of web-based technical documentations

A new framework for document verification is presented which covers the entire process from document analysis through information extraction, document modeling, representation of background knowledge about the domain of discourse, user level and formal representation of consistency criteria, verification by model checking, counterexample generation, and error reporting. Emphasis is placed on employing background knowledge to reduce the complexity and to increase the quality of results in each step. A rule-based approach to information extraction supports the concise definition of extraction rules for document formats based on XML or HTML. The expressiveness of the existing extraction methods is exceeded by supporting rule specialization, integration of external tools, and access to background knowledge represented in ontologies. As a formal basis for representing consistency criteria, the new temporal description logic ALCCTL is proposed. In contrast to the existing formalisms, criteria related to the coherence of content along individual paths of reading can be represented and verified efficiently. The adequacy, performance, and effectiveness of the proposed framework is demonstrated on a case study in technical documentation.

From Counterexamples to Incremental Interactive Tracing of Errors

Abstract This article summarizes the goals and the first results of the project CITE — from Counterexamples to Interactive Incremental Tracing of Errors — at the National Institute of Informatics (NII) in Tokyo, Japan. CITE aims at developing the fundamental methods for generating both comprehensive and comprehensible error reports based on model checking results. Model checking is a powerful and efficient method for finding flaws in hardware designs, object-oriented software, business processes, and hypermedia applications. One remaining major obstacle to a broader application of model checking is its limited usability for non-experts. It requires much effort and insight to determine the root cause of errors from counterexamples generated by model checkers in the case of a specification violation. The CITE project addresses this problem by proposing tree structured error reports that can be refined successively to the desired level of detail according to the user´s interest. First evaluations demonstrate that the proposed approach reveals more errors and explains the cause of errors more accurately than the counterexamples of existing model checkers. Zusammenfassung Dieser Beitrag gibt einen Überblick über die Ziele und die ersten Ergebnisse des Projekts CITE — from Counterexamples to Interactive Incremental Tracing of Errors — am National Institute of Informatics (NII) in Tokyo, Japan. Ziel ist die Generierung von präzisen, vollständigen und dennoch verständlichen Fehleranalysen mit Hilfe von Model-Checking. Model-Checking ist geeignet, um manuell schwer zu findende Fehler in komplexen Hardware- und Softwaresystemen automatisch nachzuweisen. Eine bestehende Hürde für die breitere Anwendung von Model-Checking, etwa zur Fehleranalyse von Geschäftsprozessen und Web-Anwendungen, ist die eingeschränkte Benutzerunterstützung. Die durch Model-Checking generierten Fehlerberichte sind oft lang und schwer zu interpretieren. Wir schlagen eine neue Methode für die Generierung von hierarchisch strukturierten Fehlerberichten vor, die in Interaktion mit dem Benutzer schrittweise verfeinert werden können. Erste Experimente im Anwendungsgebiet von Web-Dokumenten bestätigen, dass das vorgeschlagene Verfahren geeignet ist, um mehr Fehler zu finden, die Fehler präziser zu lokalisieren und die Fehlerursache besser nachvollziehbar zu machen als dies mit bisherigen Model-Checking-Verfahren möglich ist.

Checking content consistency of integrated web documents

A conceptual framework for the specification and verification of constraints on the content and narrative structure of documents is proposed. As a specification formalism, CTL

Context Aware Reuse of Learning Resources

_{\cal {DL}}