K Bertels

A cognitive approach to program understanding

A method of program understanding is presented, based on a cognitive model of programming knowledge, and incorporating the basic concepts, structures, and strategies used by human expert programmers. The approach involves the generation of a high-level description of the program with, as main advantages, the use of abstraction and robustness with respect to conceptual errors. The use of abstraction permits the transcending of any syntactical variation and allows focusing on the semantics. This also potentially allows the analysis of larger programs. The method can cope with erroneous code, and can generate a precise description of the bug(s) involved. The method is useful for integration in a tutoring system for programming.<<ETX>>

A Cognitive Model of Programming Knowledge for Procedural Languages

Recently many studies focusing on diverse aspects of programming knowledge were published. However, none of them proposes a complete model in which the contents and overall structure are defined. One of the first (and only) attempts to define such a model was made by Shneiderman, where he defines Programming Knowledge to contain two distinct knowledge bodies. We complement this model by defining the link between the syntactic and semantic knowledge. This link is constituted by the Semantically Augmented Programming Primitives (SAPPs). The paper is concerned in the definition of this ‘missing link’ in programming knowledge.