RAMOS (ID:7301/ram004)Rule-based graphcial analysis systemfor Recovery and Manipulation of Situations Robin Stanton's heuristic approach to graphical description Initially at Department of Electronic Computation, University of New South Wales University of New South Wales then at Canberra Division of Computing Research, CSIRO as part of the VERBIGRAPHICS research group including Marin Clowes Based on the ring structures in Sketchpad and Coral, but with a gestural basis; written in PLEXPACK Constraint system itself based in part on Logical Theorist Related languages
References: in Kaneff, S. (ed) Picture Language Machines: Proceedings of a Conference held at the Australian National University, Canberra on 24-28 February, 1969 view details in Kaneff, S. (ed) Picture Language Machines: Proceedings of a Conference held at the Australian National University, Canberra on 24-28 February, 1969 view details Research into the interpretation of graphics has been motivated primarily from two sources. The first is simply the desire to realise in man-machine communication the kind of descriptive power supported by the use of graphics in man-man communication. The second springs from working with a data base of information which is most conveniently recorded in graphical form (e.g. maps, engineering drawings, etc.). The quality associated with this kind of interpretation is captured by the idea of the 'machine perception of graphics'. The acceptance of this idea places computer graphics and with it, graphic languages, in a cognate position with respect to picture interpretation and scene analysis, although there are, of course, important differences. The body of the paper is concerned with reviewing the status of graphical languages given that the task for which they have to be suited is one of interpretation. Extract: RAMOS In Stanton (1970) a language, RAMOS, for recovering the interpretations of graphics is described. The language derives from the idea that graphics represent. That is, that graphical structure and problem structure are bound together under a relation of representation. The basic data types are objects, together under a relation of representation. The basic data types arc objects, attributes, relationships and sets. A complex of interrelated objects and their attributes is called a situation. For example, a situation described as "a line having an end point inside a circle" could be expressed as (Ll:LlNE L2.END(LI) L3:CIRCLE INSIDE(L2,L3)) where Ll, L2, L3 are labels, LINE and CIRCLE are objects, INSIDE is a rclationship and END is an attribute having as value a POINT object. The rules allow new entities to be described in terms of situations and a set of directives. The rule form is as follows: NAME D1 (SITUATION) D2 (SITUATION). . . . Dn (SITUATION) The directives (D's) allow descriptions to be built up in terms of procedural notions. They are selected from the following set: 1. G - (Given) the situation has been 'passed' to the rule, i.e. it is a known context 2. F - (Find) find a particular instance of the situation 3. NF - (Not Find) the situation must not be one which can be found 4. FA - (Find All) find all instances of the situation 5. FO - (Find Once) find any instance of the situation (i.e. all possible instances have the same status so there is no need to look for an alternative) 6. C - (Create) create the situatjon in the data base 7. R - (Return) return the situation as the value of the rule The system has three components; the set of rules, the data base and the acceptor. The data base is a ring based plex structure comprised of components which directly represent objects, attributes and their values, sets and relationships. Data structure is formed by using the create directivp, effectively putting structure assignment and choice of representation under the control of the User. For example, the user can represent a line either in terms of related point objects or as an object having two end-point attributes. in Nake, F. and Rosenfeld, A. "Graphic Languages" Amsterdam: North-Holland Publishing Company 1972. view details Extract: RAMOS All of the above studies used structural descriptions of pictures, but none of them attempted to provide a general programming language with pattem recognition capabilities. Stanton [20] did provide a graphics language, Recovery and Manipulation of Situations (RAMOS), based on linguistic approaches. RAMOS rules and ESP3 patterns do have some similarities, but the philosophy behind a pattem match is essentially different. The RAMOS routines search a data structure for an object described by RAMOS rules. The ordering of objects in the data structure has no relation to the special characteristics of the picture that the data structure represents. On the other hand, ESP3 routines search a special representation of a picture to find the left topmost subpicture which matches a given ESP3 pattern. in IEEE Transactions on Software Engineering (TSE) 3(2) 1977 view details |