An important software design activity is the decomposition of complex systems into conceptually independent modules that cooperate to achieve a desired result.This modularization represents a significant software engineering activity that continues to receive considerable research attention.This paper illustrates how software may be modularized by automatically determining the cohesiveness of modules in the system.Module cohesion is defined to be a quality attribute that seeks to measure the singleness of purpose of a module.We propose a metric that measures the cohesion of individual subprograms of a software system as related to each other.This metric is illustrated with detailed examples and is supported with empirical evidence supporting the viability of the measure.

It is shown how the semantic basis underlying modern system structure constructs (supported in languages such as Ada and C++) can be used to simplify the understanding of software written in conventional imperative languages. The restructuring technique analyzes the input source code system into a hierarchical system structure that exploits information hiding and localization properties. This hierarchical view will help to accurately identify a component's dependencies and visibility relationships with other system subcomponents. Besides obvious understandability and maintainability benefits, this form of restructure offers a convenient framework for translating conventional languages into languages that support modularity, abstract data types, and hierarchical system structure. The current restructuring tool has been successfully applied to an antisubmarine warfare subsystem.>

Truth maintenance systems (TMSs) offer the most elaborate approach to the belief revision problem and are thus ideal candidates for automating belief revision during the course of plausibility-driven design evolution. However, the theory of plausible design (TPD) necessitates multivalued belief propagation while conventional TMSs are two-valued (IN/OUT) systems. Due to this mismatch, the encoding of constraint dependency graphs (CDGs) in conventional TMSs requires an exponential number of justifications. A belief revision system for TPD called the theory of plausible design belief revision system (TPD-BRS) is proposed. TPD-BRS uses a modified version of the assumption-based TMS label propagation algorithm in conjunction with a label interpreter to support automated belief revision in TPD efficiently.>