Ulrich Güntzer

On the evaluation of recursion in (deductive) database systems by efficient differential fixpoint iteration

Based on matured database technology the paper provides new insights into efficient ways to evaluate recursive deduction rules. We show how the forward-chaining approach to deduction can flexibly be married with goal-directed aspects of best/easiest-first strategies. From the natural fixpoint semantics of recursion we develop generally applicable differential iteration schemes that efficiently compute the fixpoint. Surprisingly the well-known Warshall-algorithm gets disclosed as a descendant of this class of algorithms. Performance measurements suggest the former as well as systolic Δ-algorithms with linear fixpoint equation as candidates for incorporating a transitive closure operator in databases. As a next important step towards integration of database technology and logic programming we suggest to profit from the standard features of concurrency control and transaction management by effectively using them for the synchronization of parallel deductions.

Automatic thesaurus construction by machine learning from retrieval sessions

Abstract Users of information retrieval systems (IRS) know and use many relationships between concepts a long time before these find their way into textbooks, printed thesauri, or classification schemes. We present here an IRS component called TEGEN, which taps this expertise by automatically drawing conclusions from actual search behavior about possible thesaurus entries. This is done during an iterative knowledge acquisition process: only after explicit or implicit confirmation by other users of the IRS during the knowledge verification process, the results are incorporated into a thesaurus. TEGEN is written in PASCAL using a knowledge-based programming method. It uses the relational database system IMF2 and is implemented at the Technical University of Munich and at the Leibniz Computer Center of the Bavarian Academy of Sciences.

Combining Deduction by Certainty with the Power of Magic

After about two decades since the origin of relational database technology its benefits and shortcomings are well recognized in academics and in the market place. Many efforts have been focused during the past five years on extending current relational database technology towards deductive databases which offer a more powerful reasoning capability based on Horn clause logic; i.e. perform exact reasoning. Due to the ubiquity of uncertain knowledge in real-life applications, the integration of uncertainty in deductive databases is highly desirable, too, and led to so-called quantitative deductive databases very recently. As this integration can be achieved in a natural way preserving the foundations of fixpoint semantics and relational algebra, established query optimization techniques can be taken over directly. This paper focuses on how to integrate the popular magic methods for optimizing recursion in the quantitative deductive database framework. We describe two refined methods, the QMagic Set method and the Supplementary QMagic Set method. Our performance results for a series of benchmarks demonstrate their increased efficiency compared to the standard magic methods. As a specialty we show how to efficiently implement certainty-guided deduction by Yo-Yo iteration.

Time optimal left to right construction of position trees

SummaryIn the following paper we are presenting a new algorithm for the on-line construction of position trees. Reading a given input string from left to right we are generating its position tree with the aid of the general concept of infix trees. An additional chain structure within the trees, called tail node connection, enables us to construct the tree within the best possible time (proportional to the number of nodes).

DBA*: Solving Combinatorial Problems with Deductive Databases

The evolution of database (DB) technology, with its origins in hierarchical databases, has currently reached the stage of matured relational DB systems and is about to grow into deductive DB systems, where logic programming — originated by the artificial intelligence (AI) community — plays a central theoretical role. Until now, however, no smooth integration of another important part of AI, namely that of heuristic search and intelligent planning, into DB technology was known. This paper contributes a first step beyond deductive DB systems towards intelligent DB systems. We describe the well-known A*-algorithm in terms of a general theoretical framework for deductive DB systems, the sloppy deltaiteration scheme, and give a generalized algorithm, called DBA*-algorithm. As an immediate consequence, heuristic search strategies for combinatorial problems become now feasible in the DB environment in a natural and efficient way. We also present a prototype implementation of the DBA*-algorithm, with the 15-Puzzle and the Traveling Salesman Problem as sample combinatorial problems. The benchmark results gained from this testbed demonstrate the applicability and efficiency of our approach for heuristic search in deductive DB systems.

Semantics and efficient compilation for quantitative deductive databases

A coherent approach is presented that extends relational and deductive database technology toward an integration of expert-system applications, which require sound and efficient capabilities to deal with uncertainty. Extending logic programming, the authors define the semantics of quantitative deductive databases, where fixpoint theory plays a central role. Calculus gives the rule programmer a great deal of flexibility to tailor the aggregation of certainties according to the application expertise at hand. Extending relational algebra, the authors also introduce a quantitative relational algebra as a suitable target language for rule compilation. This approach makes rule-based expert systems requiring uncertainty reasoning on large and complex data, feasible for a variety of practical application areas.<<ETX>>

Ein Expertensystem zur Unterstützung der Filialanalyse bei der HYPO-Bank

Im vorliegenden Beitrag wird ein Prototyp-Expertensystem zur Unterstutzung der Filialanalyse bei der HYPO-Bank vorgestellt.