Tackling Uncertainty in A Cognitive Framework for Source Code Understanding

Abstract

Source code understanding plays an important role in software maintenance. Although existing knowledge-based program understanding techinques distinguish themselves from structural/syntax-based program analysis methods for acquiring high-abstraction-level oriented result, they all fail to address uncertainty issues inherited in program comprehension process. The lost benefit from tackling huge amount of uncertainty information renders those knowledge-base methods both ineffective and inefficient; in most cases, they simply provide a framework where the linking among different information sources is manually arried out by programmers. In this paper, we devote ourselves to addressing the uncertainty issues in a cognitive framework proposed in previous work for code understanding [12]. An example is given to illustrates this approach.