The RESEDA project was formed to provide specialists of the XIVth and XVth centuries with an interactive method of processing biographical data. At present, thanks to the joint financing of the Centre National de la Recherche Scientifique and the D616gation G6nbrale A la Recherche Scientifique et Technique (contract 75.7.0456), a test model is being worked out with the leading characters of the humanist literary movement in France. The methodological interest of the project seems to lie essentially in the attempt to enrich the techniques of information retrieval by instruments endowed with a certain deductive power (Cros et al. 1968; Fossier, Cr6hange 1970) in order to assess, by working with a concrete example, the tools necessary for simulation of quite complex intellectual processes. Moreover, owing to the extremely complex and very often implicit nature of the relations between the characters who interest us, it is almost essential to have recourse to very sophisticated methods. The target is, in fact, the creation of a system which would have not only a static function of recuperating a stored item of information but also a dynamic function of establishing new links between the data and increasing, as it were, the initial knowledge. Therefore RESEDA must to a certain extent enter the sphere of artificial intelligence. I ought to put aside right away any ambiguity about the way we intend to use this new discipline. Our approach is quite different from that of Schank, for example (Schank 1973; Schank et al. 1975) or of Minsky (Minsky 1974) to mention but two of the most famous names. Schank wanted to set up a theory of the conceptualization phenomena which underlie the surface structures of natural language in order to be able to understand it fully by automatic means. As for Minsky, his frames system was created to adapt itself to any given situation, in its perception as well as in its linguistic formulation, and it remains at a level of generality which allows it to take into account all the aspects of a certain reality. RESEDA does not concern itself at all with such problematic and very carefully avoids general systems; unlike the above-mentioned systems, formulating hypotheses about the workings of human thought. So all our references to any theory based on a certain analysis of mental processes (see for example the correlators theory of Ceccato, Ceccato 1962) spring only from a pragmatic interest in a useful tool, and never from adherence or adoption of a definite position about theoretical foundations. RESEDA, well placed in its humanist context, does not set itself any problem of "weltanschauung." In order to reach our goal, our borrowings from artificial intelligence are, for the description of basic data, methods of the case systems type which are very rich in deductive possibilities and free from formal faithfulness to expressions in natural language (Schank 1973, 1975; Bruce 1975); to use this descriptive richness, we must develop an inference system based either on principles like the notion of analogy, transitivity etc. (Schwarcz et al. 1970; Carbonnel, Collins 1974), or more ambitiously, on taking into account all the constituent properties of certain given situations (Charniak 1972, 1975, 1976; Minsky 1974; Schank et al. 1975).
Read full abstract