View source for User talk:Jean-Philippe de Lespinay/La Maieutique

La Maieutique<ref>[http://www.automatesintelligents.com/echanges/2008/dec/ialespinay.html Jean-Paul Baquiast and Christophe Jacquemin], Automates Intelligents, "Dossier: The Reasoning Artificial Intelligence" (2009): "[http://www.automatesintelligents.com/echanges/2008/dec/iajpl.pdf Genesis of an invention: La Maieutique]"</ref> is the knowledge engineering method invented by [[Jean-Philippe de Lespinay]] in 1986. It showed its effectiveness the same year, producing the expert system ''Josephine'' (1000 rules and several external programs) which surprised the business world at the time<ref>[http://www.tree-logic.com/Articles/01%20Hebdo%20et%20Jos%C3%A9phine%20et%20Avignon%2087.pdf 01 Informatique], A system in gold to talk about money, April 21, 1987</ref><ref>[http://www.tree-logic.com/Articles/Le%20Point%20&%20Nouvel%20Economiste.jpg Le Point] February 2, 1987: "Bank : artificial money"</ref><ref>[http://www.tree-logic.com/Articles/Josephine-Les-Echos.jpg Stanislas du Guerny], Les Echos, Computer decides for you at the Banque de Bretagne, January 28, 1987: "</ref><ref>[http://www.tree-logic.com/Articles/Le%20Point%20&%20Nouvel%20Economiste.jpg Le Point and Le Nouvel Economiste] about Josephine expert system, February 1987</ref><ref>[http://www.tree-logic.com/Articles/Bancatique.jpg Olivier Chedeville], Bancatique, Expert system for financial investment advice, a first at the Banque de Bretagne, February 1987</ref>. This software was developed by a former bank employee with virtually no knowledge of computing and Artificial Intelligence: Michel Le Séac'h, who reported this experience, and more, in a book<ref>Le Séac'h, Michel, [http://www.sudoc.abes.fr/xslt/DB=2.1/SRCH?IKT=12&TRM=021351031 Développer un système expert] (How to Develop an Expert System), PSI, Paris 1989</ref>. An industry first because at this time - as today in 2011 - development of experts systemes was a complex, long and expensive procedure requiring whole teams of IT specialists<ref>[http://books.google.com/books?id=WLRTdGxJMrUC&pg=PA448&dq=%22maintenances+adaptative+et+curative%22&hl=fr&ei=OJl4TtXrDa7S4QS5r5DoCw&sa=X&oi=book_result&ct=result&resnum=1&sqi=2&ved=0CDUQ6AEwAA#v=onepage&q=%22maintenances%20adaptative%20et%20curative%22&f=false Kenneth Laudon, Jane Laudon, Eric Fimbel, "Management Information Systems: Managing the Digital Firm", Business & Economics, 2010 edition, chapter 11-3.5] : ''The implementation of a large number of expert systems requires the deployment of considerable development efforts, lengthy and expensive. Hiring and training a larger number of experts may be less expensive than building an expert system .(...) Some expert systems, particularly the largest, are so complex that over years, the costs of curative and adaptive maintenance become as high as the cost of development''</ref>. Other innovation, Josephine dialogued with the end user on-premise. As it was installed in bank agencies to be available to clients, journalists could test it incognito, which leads to some bad articles in the press<ref>[http://www.tree-logic.com/Articles/Joséphine%20fera%20date.jpg Le Monde Informatique], May 30, 1988: "A try to convert at Bank of Brittany - Is Josephine operational ? Yes, but ..."</ref>

It allows to program without computer training, developing in fact the program as an expert system, since it is easier to develop an expert system than a classical program, provided it has a reasoning engine (zeroth order engine)<ref>Science et Vie Article about Artificial Intelligence state of art, Jean-Philippe de Lespinay, May 1991, [http://www.tree-logic.com/scienceetvie.htm From total zero to Zero Plus]: "Artificial Intelligence has really given birth, and more in France, a remarkable tool: the second generation expert system (known as 0+ order ), endowed with a faculty of human reasoning and with an extraordinary ease of use"</ref>. Any program is divided into three parts that must be programmed : an expertise (or a knowledge), [Data|data]] and human-machine interfaces.

There are user friendly]] tools in the market to manage data (Excel, SQL, online tools, etc.)<ref>[http://www.ncbi.nlm.nih.gov/pubmed/21718534 Pandey RV, Nolte V, Boenigk J, Schlötterer C.], University of Vienna, CANGS DB, a stand-alone web-based database tool for processing, managing and analyzing 454 data in biodiversity studies: "''Being a local database that is accessible through a user-friendly interface, CANGS DB provides the perfect tool for collaborative amplicon based biodiversity surveys without requiring prior bioinformatics skills''"</ref> and interfaces<ref>[http://www.tripwiremagazine.com/2011/03/50-user-interface-design-tools-a-web-designer-must-have.html Sonny M. Day],Tripwire Magazine, March 10, 2011, User Interface Design Tools A Web Designer Must Have</ref>. The non computer scientist developer doesn't need to develop again that two parts. By cons, the expertise automation is a very difficult task in programming and, more, in software maintenance<ref>[http://www.pitt.edu/~ckemerer/CK%20research%20papers/SwComplexityAndMaintenanceCost_BankerDatar93.pdf Rajiv D. Banker, Srikant M. Datar, Chris F. Kemerer and Dani Zweig], Software complexity and maintenance costs</ref>. Any method to accelerate these two phases is welcome. Precisely the collection of knowledge is the specialty of expert systems. By combining the knowledge automation with La Maieutique, an interface generator and a database tool, Jean-Philippe de Lespinay said he obtained the equivalent of a programming language. For him, La Maieutique was also a method of conventional program developing.

The process of La Maieutique is original because it does not ask experts to describe the knowledge to put in the program, they are generally not able. It asks them to tell "how they do" to solve the same problem that the program should do. This is very easy for them and produces decision trees in the expert language<ref>[http://www.tree-logic.com/Articles/arbre%20paie%20T.Rex.pdf Maieutica decision trees] (example of a screen copy)</ref>, representing his know-how. La Maieutique extracts from these trees the underlying knowledge and expresses it in the form of expert system rules<ref>[http://www.tree-logic.com/Articles/Règles%20Maieutica.JPG Maieutica rules] (example of a screen copy)</ref>,<ref>[http://www.tree-logic.com/Articles/Maieutica%20avec%20explication.JPG Maieutica reasoning explanation] (example of a screen copy)</ref>. With a reasoning engine, expert system is working from the generation of the first rule. Experts can thus control developing from the first minutes of their interview.

From 1986 to 1991, this method was manual. Developers drew the trees and extracted the rules. Josephine and Createst were developed like that. From 1991 with the birth of Maieutica, expert interview was guided by software, which itself drew trees, extracted rules and run the expert system, all in one operation<ref>[http://www.tree-logic.com/Articles/Dialogue%20PC%20diagnosis%20anglais.jpg Maieutica interfaces] (example of a screen copy)</ref><ref>[http://www.tree-logic.com/Articles/English%20rules.jpg Rules generated by Maieutica] (example of a screen copy)</ref>. Hundreds of expert systems were developed in this way. The software produced was naturally conversational, asking the right questions at the right time what a classical program can not do. To integrate the expert system in a conventional application, it was necessary to produce interface with a standard interface generator or a computer language. To manage a database, it was necessary to use a database manager or a computer language.

In 2001, Maieutica was replaced by T. Rex (Tree Rules EXtractor), a software generator containing the needed interfaces to allow a not computer scientist to develop himself a program (as an expert systems), to integrate it in applications and to manage database by reasoning, all without programming language.

== References ==
{{Reflist|1}}