Hypothesis-based concept assignment to support software maintenance

Gold, Nicolas

doi:10.1109/icsm.2001.972768

Cited by 21 publications

(19 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our approach bears some similarity to that of Gold [8] who designed a system (HB-CA) to solve the concept assignment problem. His system uses a knowledge base of programming concepts that play the same role as the business domain model in our work.…”

Section: Related Workmentioning

confidence: 99%

A three-layer model of source code comprehension

Belmonte

Dugerdil

Agrawal

2014

Proceedings of the 7th India Software Engineering Conference

View full text Add to dashboard Cite

In this paper we first propose a source code comprehension model built as a hierarchy of three abstraction levels from the source code to the purpose (goal) of the program. The elements belonging to each layer have been precisely defined as well as their links to the elements in the adjacent layers. Consequently this model allows to bridge the semantic gap between the purpose of the program defined in business terms and the code that implements it. The model leverages two ontologies: an action ontology, which is specific to our approach, and a domain concept ontology. Next this model has been implemented as a tool under Eclipse and two experiments have been performed to assess the relevance of our approach in the maintenance of a large-scale program. The results of this experiment are very encouraging. The contribution of the paper is the presentation of our program comprehension model built on a novel approach based on an action ontology, the description of the tool we developed to assess the relevance of model and the testing of the latter with two controlled experiments.

show abstract

Section: Related Workmentioning

confidence: 99%

A three-layer model of source code comprehension

Belmonte

Dugerdil

Agrawal

2014

Proceedings of the 7th India Software Engineering Conference

View full text Add to dashboard Cite

show abstract

“…The data per location of concept bindings are same as those in previous chapter which were identified by the WeSCA tool [51]. The CodeSurfer [53] tool's API is used to identify principal variables in analysis scopes and to implement the KSA algorithm to find key statements.…”

Section: Analysis Toolsmentioning

confidence: 99%

“…Gold [50] introduced the Hypothesis-Based Concept Assignment (HB-CA) to address the concept assignment problem. This method is computationally inexpensive, having linear computational growth in the length of the source code under analysis.…”

Section: High-level Analysis Based Techniques For Program Comprehensionmentioning

confidence: 99%

Identifying high-level dependence structures using slice-based dependence analysis

2009

2009 IEEE International Conference on Software Maintenance

View full text Add to dashboard Cite

This thesis addresses dependence based concept assignment using slicing based dependence analysis on concept assignment.Concepts are domain level entities concerned with the real world. Although concepts are defined at the domain level, the source code of a program must, nonetheless, express these concepts in program code denotations. Traditional concept assignment maps the low level of source code to the high level of the domain through the medium of associations. These associations typically take the form of comments and identifiers.Previous work on the combination of dependence analysis with concept assignment gave the definition of either extending concepts to concept slices or abbreviating concepts to key statements. However, there are no empirical studies to show the feasibility of these two approaches.In this thesis, we propose an improved key statement analysis algorithm and empirically study executable concept slicing and key statement analysis using ten subject programs with more than 600 concept bindings identified. We define dependence based metrics to evaluate both approaches and to provide evidence of both advantages and disadvantages. Based on these foundations, we propose a new combination technique -concept refinement, which adopts the advantages and overcome the disadvantages of previous two approaches.The results for the same programs with the same concept bindings showed that concept refinement can produce dependence based concept bindings. The refined concept binding is smaller, but with similar impact and higher cohesion than the original HB-CA concept binding. Therefore, it becomes more precise after refinement; the refined concept binding is much smaller than the executable concept slice with respect to the original HB-CA concept binding. These findings suggest that the refined concept binding is more reasonable in size for program comprehension.Furthermore, the refined concept binding can be extracted from most of the original HB-CA concept bindings, thus concept refinement is more appropriate than the key statement analysis. DeclarationThe work presented in this thesis is original work undertaken between September

show abstract

“…There are many types of tools available to help with software comprehension, emphasising different aspects of systems and modules, and usually creating new representations for them [10].…”

Section: Comprehension Supportmentioning

confidence: 99%

“…Examples of this type include DM-TAO [3], [4], IRENE [17], and HB-CA [10], [12]. One exception to this categorisation is Hartman's work [14] that falls between approaches 2 and 3.…”

Section: Comprehension Supportmentioning

confidence: 99%