fuchs@frederic.ifi.unizh.ch (11/28/90)
Everynow and then I get into a situation where I have to argue for Prolog's practicability. It would help to point to success stories -- successful products based on Prolog, visible advantages of using Prolog as a development tool, e.g. for executable specifications, remarkably better quality of software products based on Prolog or developed with the help of Prolog, etc. I know of some success stories, e.g. the German PTT's XTEL, an expert system for the configuration of telephone equipment, or Sandoz' scheduling system for chemical production, but it would certainly help to know more, especially to kmow why people decided to use Prolog, which advantages they expected, which results they got. I remember that Miguel Calejo of the University of Lisbon asked some time ago a similar question. Unfortunately, I never saw all the answers. Please e-mail to me directly. I will submit a summary to the net. Norbert E. Fuchs Department of Computer Science University of Zurich CH-8057 Zurich fuchs@ifi.unizh.ch
fuchs@frederic.ifi.unizh.ch (Norbert E. Fuchs) (12/29/90)
In article <1990Nov28.095835.4265@ifi.unizh.ch> I wrote: >It would help to point to success stories -- successful products based on >Prolog, visible advantages of using Prolog as a development tool, e.g. for >executable specifications, remarkably better quality of software products >based on Prolog or developed with the help of Prolog, etc. In the meantime I received the following messages. I would like to thank all who answered my request. --- nef -------------------------------------------------------------------------------- Expertech's Xi Plus Expert Systems shell was originally written largely in prolog. Unfortunately, it may not be a very positive example, as it has now been rewritten into C to speed it up (though this may be more related to the quality of the prolog interpreters that were available on the PC at the time than anything else) Bob McKay <rim@csadfa.cs.adfa.oz.au> -------------------------------------------------------------------------------- We've used Arity PROLOG to develop a Papaya Management System for the PC environment that works with 4 areas of expertise. PROPA: a papaya management expert system Stephen Y. Itoga, C. Leng Chia, Russel S Yost, and Ronald F L Mau, Knowledge-Based Systems, vol 3, no 3 Steptember 1990, pp 163-169. Stephen Itoga <itoga@uhccux.uhcc.Hawaii.edu> -------------------------------------------------------------------------------- You might know about this already, but the MIT Press book _The Practice of Prolog_ consists of case studies of Prolog programming in the "real world". I don't know how many, or in what depth they are treated, but it might at least give some pointers. It's in the same series as _The Art of Prolog_ and _The Craft of Prolog_. Jamie Andrews <jha@lfcs.edinburgh.ac.uk> -------------------------------------------------------------------------------- Here at Unisys, Prolog has been used on some bread-and-butter projects. We've used it to develop a very large configuration expert system (described in a recent paper in JLP) and more recently to develop an expert system to assist the Unisys spare-parts inventory management department (described in a paper to appear in CAIA-91) and in a very large expert system to diagnose problems in large letter-sorting machines used by the US Postal system (described in various conference papers). Tim <finin@PRC.Unisys.com> -------------------------------------------------------------------------------- Check out Journal of Logic Programming vol. 8 no. 1-2, Jan. 1990 -- this is a special issue on Applications. Saumya K. Debray <debray@cs.arizona.edu> This issue contains the following papers ----------------------------------------- Aunt: A Universal Netlist Translator (P. B. Reintjes) Python: An Expert Squeezer [for the game of Bridge] (L. Sterling, Y. Nygate) Modeling Simultaneous Events With Default Reasoning (A. Van Gelder) Logic-Based Configuration With a Semantic Network (D. B. Searls, L. M. Norton) Solving Large Combinatorial Problems In Logic Programming (M. Dincbas, H. Simonis, P. Van Hentenryck) Mockingbird: A Logical Method For Testing (M. M. Gorlick, C. F. Kesselman, D. A. Marotta, D. S. Parker) Logic Programming For Real-Time Control of Telecommunication Switching Systems (N. A. Elshiewy) An Expert System For Harmonizing Chorales In the Style Of J. S. Bach (K. Ebcioglu) -------------------------------------------------------------------------------- You might try to contact David Miller david@itc.icl.ie. They have been using the SEPIA/CHIP system for various purposes, some of them really spectacular. But I cannot guarantee that he wants to be public about them :-) Micha Meier <micha@ecrc.de> -------------------------------------------------------------------------------- Recently you posted to comp.lang.prolog, asking for stories about Prolog's practicality. You could mention Electronique Serge Dassault -- they use Prolog a lot (in fact, they have written an object-oriented workbench, which IBM sells). You might be able to get more details from your local IBM office; I would tell you more but I am about to go on vacation and am rather busy right now. There's also the book: Lazerev, G.L.: "Why Prolog: Justifying Logic Programming for Practical Applications" (Prentice-Hall 1989) Peter Ludemann <ludemann@mlpvm1.iinus1.ibm.com> -------------------------------------------------------------------------------- ... a couple of months ago we got a visit from"customers" of BNR Prolog (who have done some interesting experiments with interval arithmetic): Dassault Electronique. They claim to be one of the biggest industrial users of Prolog, in Europe anyway (probably Siemens is the biggest). One of their commercial products, developed entirely in Prolog is a VLSI design tool called FrenchChip. It takes a specification in VHDL and synthesizes your ASIC. They claim (in their glossy brochure) that it "takes care of *testability*, inter-blocks loading and buffering, timing along critical paths, gate and component counts". If you want to talk to someone at Dassault Electronique directly, feel free to call Patrick Taillibert 33 (1) 34 81 60 28. I don't think they have e-mail. As for Prolog at BNR, our reasons are the usual: rapid prototyping for small-medium sized projects, compactness of code, etc. We do seem to have the phenomenon of religious factioning, however. For some reason, people either love Prolog or hate it. My explanation for this is that programers who are used to procedural programming suddenly feel like they have lost control of what their program is doing because they haven't told it what to do explicitly. Consequently they have no confidence that it does what they intend. They grumble about unwanted backtracking and lack of type-checking and so on but the truth of it is that the learning curve is just a little steeper than Pascal. I suspect that they don't really understand the language. Having said that, I think there is a lot of room for Prolog to improve. We need a next-generation logic programming language, but I have no idea what it should look like. Andre' Vellino <vellino@bnr.ca> -------------------------------------------------------------------------------- I wrote an intelligent back-end to a farm management package. The thing modelled pigs growing in a pig pen. My system (called "pigE") inputted screens and screens of numbers and outputted one page of text recommendations. We choose prolog because: 1) we had a very limited time to do a prototype and prolog was my tool of greatest productivity 2) the thing was to be marketted at less than $2000. It already had in it several third-party packages that would all require liscencing fees in the final system. The users were very anxious not to incur any more liscening fees, so they wanted a **langauge** not a package. The thing was written originally under DOS in UNSW Prolog, then taken to Arity Prolog (cause it had a good windowing system). One of the cute features of the system was a rule langauge customised to describing the modelling package. In the end, the domain experts looked after the knowledge base. The whole system (modelling package + Prolog expert system) was subsequently sold to BP Purina (a US argricultural nutrition company) for $2.5M. The OZ developers were put on a (very healthy) retainer for that period. Reference: "Combining Heuristics and Simulation Models, An Expert System for the Optimal Management of Pigs", Proceedings of the1988 Australian Joint Artificial Intelligence Conference. <timm@runxtsa.runx.oz.au> -------------------------------------------------------------------------------- The following three applications were written in Prolog SECReTS, an expert system for credit evaluation written by TecLogic in Padova, Italy. A prototype for a Production Planning and Scheduling System developed by Sandoz International in Basle, Switzerland. XTEL: a configuration expert system for telephone equipment for the German PTT developed by Interface Computer Munich, Germany. --- nef
mcovingt@athena.cs.uga.edu (Michael A. Covington) (01/07/91)
Is _The Practice of Prolog_ (sequel to The art... and The craft...) available now? If not, when will it be out?