Ulrich Geske

Web Knowledge Management and Decision Support

Shared declarative aspects of Prolog and XML are examined. An XML version of pure Prolog is shown to be at the center of the Rule Markup Language. The RuleML data model uses Order-Labeled trees, combining the RDF and XML models. As part of RuleML’s hierarchy of sublanguages, the RuleML-Prolog DTD is developed into an XML Schema. XSLT (XSL Transformations) is employed for practical XMLto-XML and XML-to-(X)HTML transformation of Prolog on the Web.

Railway scheduling with declarative constraint programming

Simulation of train scheduling is a highly complex problem. Classical methods in this field are mainly designed for conflict resolution, which means that a solution or partial solution is generated and subsequently tested to determine whether the conditions are met (generate-and-test procedure). The main advantage of the proposed paradigm, Constraint Processing, is that its basic strategy is avoidance of conflicts. The use of the conflict-avoiding CP paradigm is advantageous, for example, in scheduling trains (track selection, global temporal situations, reservations), where strongly branched decision trees arise. Some examples are given illustrating the innovative aspects of the Constraint Processing paradigm. However, the size of real problems, in terms of track length, number and type of trains, different disposition rules, optimization or quality criteria, make it necessary to explore other methods to deal with the amount of data, to reduce the remaining search spaces, to ensure short response times and interactivity and to guarantee high-quality solutions. We describe possible ways of coping with the above mentioned problems, especially to reducing the lateness of trains: automatic decomposition of large rail networks and distributed train scheduling, using a slice technique to improve the systems backtracking behaviour with a view to finding faster, better solutions, and combining constraint processing and genetic algorithms to find alternative tracks in a station.

Constraint-based configuration of large systems

Configuration processes in industry are complex and until now have been time-consuming and costly, even though computers are sometimes used. Furthermore, the configuration found is seldom an optimal or near-optimal solution. We developed a constraint-based system for expert driven configuration that yields good solutions in fairly quickly time depending on the information and the expert knowledge available. We have now extended this approach to deal with large systems consisting of hundred or thousands of components. We describe methods to limit memory-consuming processes and provide results for benchmark problems.

A Guide for Manual Construction of Difference-List Procedures

The difference-list technique is an effective method for extending lists to the right without using the append/3 procedure. There exist some proposals for automatic transformation of list programs into difference-list programs. However, we are interested in a construction of difference-list programs by the programmer, avoiding the need of a transformation. In [9] it was demonstrated, how left-recursive procedures with a dangling call of append/3 can be transformed into right-recursion using the unfolding technique. For some types of right-recursive procedures, the equivalence of the accumulator technique and difference-list technique was shown and rules for writing corresponding difference-list programs were given. In the present paper, improved and simplified rules are derived which substitute the formerly given ones. We can show that these rule allow us to write difference-list programs which supply result-lists that are either constructed in top-down -manner (elements in append order) or in bottom-up manner (elements in inverse order) in a simple schematic way.

Constraint-Rules for Configuration Problems

In this paper, we outline our constraint-based model for configuring industrial products and treat some aspects of its use of constrained-based rules as well as some model extensions for product reconfiguration and for the efficient solution of large configuration problems. The resulting model realizes a substantial reduction of the search space and allows the efficient configuration of industrial products. Correctness and completeness of the solution process are ensured with respect to the problem specification. Unlike most other configuration models, the proposed model offers almost unlimited user freedom to interact with the solution process.

Integrating time constraints into constraint-based configuration models

Over the last few years, we have been developing the configuration model ConBaCon, which is based on Constraint Programming over finite domains. The model is sound and suitable for building efficient and flexible systems that fulfill all the requirements of advanced configuration systems. In this paper, we present model extensions that enable time-extended configuration and reconfiguration problems to be solved: Besides “normal” configuration and reconfiguration problems, the extended model can now also solve problems that contain time-dependent resource avaibilities or supply constraints for the ground components included. The general nature of the model extensions introduced seems to be suitable for integration in other (commercial) constraint-based configuration systems/configurator libraries as well.

Automatische und interaktive Stundenplanung

Zusammenfassung.Das System f“ur die interaktive, automatische Stundenplanung ist im Rahmen der Forschungsarbeiten des Bereichs Planungstechnik und Deklarative Programmierung in Fraunhofer FIRST zur Erweiterung der Constraint-basierten Programmierung entwickelt worden. Mit dem System wird die Stundenplanung der Medizinischen Fakult“at Charité seit dem Sommersemester 1998 vorgenommen. Seitdem wurde das System kontinuierlich weiterentwickelt. Der erfolgreiche Einsatz des Systems zeigte, dass die gew“ahlten Methoden und Verfahren sehr geeignet f“ur die Behandlung derartiger Probleme sind. Die Vorteile einer kombinierten interaktiven und automatischen Stundenplanerzeugung konnten eindeutig nachgewiesen werden.Abstract.The interactive, automatic timetabling system was developed in Fraunhofer FIRST’s department for Planning Technology and and Declarative Programming in the frame of research on the constraint-based programming paradigm. Since 1998, the system has been used to generate timetables for the Charité Medical Faculty at Berlin’s Humboldt University and has undergone continuous extensions. Successful deployment of the system shows that the chosen methods and techniques are well suited for dealing with problems of this sort. The quality of the plans and their permanent use demonstrate the advantages of combined interactive and automatic generation of timetables.

Konfiguration komplexer Produkte mit Constraint-basierter Modellierung

Zusammenfassung. Industrielle Konfigurationsprozesse sind, bedingt durch die vorherrschende Variantenvielfalt, sehr komplex. Aufgrund der inhärenten Potentiale ist die Realisierung von effizienten, flexiblen Konfigurationsprozessen ein zentrales Ziel vieler Unternehmen. Wir haben einen Ansatz für die Konfiguration und Rekonfiguration von Produkten entwickelt, der auf Constraint-basierter Modellierung über endlichen Domänen basiert. Das auf diesem Ansatz beruhende System ConBaCon deckt eine breite Palette von Konfigurationsproblemen ab und ermöglicht eine effiziente Lösung, die sowohl automatisch als auch gesteuert durch hoch flexible Nutzerinteraktion erfolgen kann. Wendet man das System jedoch auf große Konfigurationsprozesse an, so treten Performanzprobleme auf. Als zwei wesentliche Modellerweiterungen, die diesem Mangel entgegenwirken, stellen wir in dieser Arbeit einerseits eine Modell-Clusterung und andererseits die Behandlung großer Netze arithmetischer Constraints vor.Abstract. The configuration of complex industrial products is often costly and inaccurate because of the numerous variants of products and their components. Improving the configuration of technical products using computer-supported methods is a prerequisite for modern, efficient and flexible customer-oriented production. We describe an approach to the configuration and reconfiguration of technical devices that is based on constraint programming over finite domains. The system ConBaCon, which uses this method, combines automatic processing with interactive user interactions. Benchmark tests are used to investigate the relations between the number of components, hierarchy depth and efficiency. Furthermore, clustering and special treatment of large networks of arithmetic constraints are presented as methods of dealing with high complexity.

Introduction to part III: decision support by advanced logic programming

In decision support systems at least two difficulties are encountered if conclusions are to be derived based on arbitrary knowledge sources. One difficulty concerns the quality and structure of the knowledge. Knowledge can be contradictory and its truth value can change over time; the representational structure of knowledge is important for the ability to access specific items of information and to derive implicit knowledge. Another difficulty concerns the design of a knowledge processing formalism; for practical reasons, this should allow fast execution, but predictable behaviour is also important.

Cardinality constraints in disjunctive deductive databases

This paper surveys work of the authors on pattern languages for tree-structured data with XML as the main application in mind. The main focus is on formalisms from formal language theory and logic. In particular, it considers attribute grammars, query automata, tree-walking automata, extensions of first-order logic, and monadic second-order logic. It investigates expressiveness as well as the complexity of query evaluation and some optimization problems. Finally, formalisms that allow comparison of attribute values are considered.