We maintain binary snapshots of the Glasgow Haskell Compiler (GHC) for Linux and HP-UX platforms. More information is in the README.
The Functional Rent Advisor (FRA) is an intelligent Internet application that allows calculating an estimate of the fair rent for a flat in Munich. The FRA is a re-implementation of the Munich Rent Advisor (technical report PMS-FB-1996-11 in our publication list) using the high-level purely functional language Haskell.
The paper The Functional Rent Advisor describes the implementation of the FRA in Haskell and shows that Haskell is appropriate to realize non-trivial Internet applications. Simplicity, flexibility and rapid prototyping were the advantages of using Haskell.
The Functional Rent Advisor is available in German and English. The version you get is chosen according to your preferred language. To choose your preferred language, set the language preferences of your browser appropriately.
Satchmo is a model generator for first-order theories implemented in Prolog. It can be seen as an interpreter, where the supplied theory is the interpreted program. To avoid this interpretation layer, the Satchmo Compiler compiles a clausal theory into a Prolog program which then generates models of the theory more efficiently.
See the paper "Efficient Model Generation by Compilation" (technical report PMS-FB-1996-2 in our publication list) for a description of the underlying ideas.
MM_SATCHMO is a generator of minimal Herbrand models for sets of first order clauses. It is a concise Prolog program in the style of SATCHMO that generates all minimal Herbrand models of the considered clauses, and each such model is generated only once.
The directory includes the MM-SATCHMO program as well as Benchmark examples.
See the paper "Positive Unit Hyper-Resolution Tableaux for Minimal Model Generation. " (PMS-FB-1997-8 in our publication list) for an explanation of MM-SATCHMO's principle
Finfimo is a model generator for first-order theories implemented in Prolog. It is an extension of Satchmo in the sense that Finfimo handles existential quantifiers. This is essential with respect to finite model generation.
See the paper "A Deduction Method Complete for Refutation and Finite Satisfiability." (PMS-FB-1998-6 in our publication list) for a description of the underlying ideas.
This is a Prolog implementation of the approach to automated model building described by C. G. Fermüller and A. Leitsch from TU Vienna.
Snarks is a graphical tool for the analysis of tableau proofs as generated by Satchmo and similar deduction systems. Snarks comes with a built-in inference engine, but it can also be hooked to external inference engines. Its user interface is implemented in Java, so it is directly runnable by following the link above.
Snarks was developed and implemented by Mathias Kettner as a diploma thesis.
INTERDIP is a constraint-based system that helps the user in generating and maintaining schedules of sanitary personnel under a complex network of constraints and preferences. InterDip is directly runnable by following the link above.
InterDip was developed and implemented by Hans Schlenker as a diploma thesis.
Der Stundenplaner erzeugt für das Institut für Informatik, jeweils ausgehend vom Stundenplan des vorhergehenden Semesters, einen neuen Stundenplan für ein Semester.
VisBenchmark is an interactive scatterplot. The visualized data is read via URLs. Each data point is associated with a URL which can be loaded with a simple mouse click.
Researchers performing benchmark experiments can use VisBenchmark to visualize the performance of two algorithms in a two dimensional scatterplot. They can select interesting data points in order to display the benchmark problems resulting in the respective data points.
VisBenchmark was developed and implemented in Java by Angelika Blauth as a Fortgeschrittenen-Praktikum.
Documentation, in German (html, ps.gz), source
Information on the Internet is often available only in the form of raw texts. Since there are only a few possibilities to format a raw text, the formatting of these texts is often ambiguous. To process the texts (e.g. in a information retrieval systems) it is often useful to have the text information separated from the formatting. For some applications (e.g. information extraction systems or information retrieval) it is even more useful to have the structure of the document available as well.
ConStruct is a system which tries to recover the structure of these documents that is expressed by their poor formatting. To achieve this goal it uses the blackboard system architecture with experts for the various different structures of texts.
Apart from the use of the ConStruct system for structuring Internet documents our system can also be used for post-processing the output of an Optical Character Recognition (OCR) system.
ConStruct was developed and implemented by Robert Reiner and Thomas Traber as a Fortgeschrittenen-Praktikum.
Categories is a tool for the interactive manipulation of classification systems represented by extended and-or trees. The system has originally been developed for an application in linguistic research, but is in no way restricted to this application area.
Categories was developed and implemented by Matthias Nickles as a Fortgeschrittenen-Praktikum.
Lehr- und Forschungseinheit
Institut
Universität