[Paper Review] Detecting commonality and variability in use-case diagram variants

Ra’Fat Al-Msie’deen, Anas H. Blasi+8 more|arXiv (Cornell University)|Mar 1, 2022

Software Engineering Research33 references10 citations

TL;DR

This paper proposes an automated approach to detect common and variable features across multiple versions of use-case diagrams using formal concept analysis (FCA) and latent semantic indexing (LSI). Applied to a mobile media case study, the method successfully identified all common and variable features with high precision, demonstrating its effectiveness in product line reengineering.

ABSTRACT

The use-case diagram is a software artifact. Thus, as with any software artifact, the use-case diagrams change across time through the software development life cycle. Therefore, several versions of the same diagram are existed at distinct times. Thus, comparing all use-case diagram variants to detect common and variable use-cases becomes one of the main challenges in the product line reengineering field. The contribution of this paper is to suggest an automatic approach to compare a collection of use-case diagram variants and detect both commonality and variability. In our work, every use-case represents a feature. The proposed approach visualizes the detected features using formal concept analysis, where common and variable features are introduced to software engineers. The proposed approach was applied on a mobile media case study to be validated. The findings confirm the importance and the performance of the suggested approach as all common and variable features were precisely detected via formal concept analysis and latent semantic indexing.

Motivation & Objective

• To address the challenge of identifying common and variable features across evolving use-case diagram variants in software product lines.
• To automate the detection of commonality and variability in use-case diagrams throughout the software development lifecycle.
• To support software engineers by visualizing detected features through formal concept analysis for improved understanding and decision-making.
• To validate the proposed approach using a real-world case study in mobile media application development.

Proposed method

• Represent each use-case as a feature and model variants as feature matrices for analysis.
• Apply latent semantic indexing (LSI) to reduce dimensionality and uncover semantic relationships among use-cases.
• Use formal concept analysis (FCA) to organize and visualize the detected features into formal concepts, distinguishing common and variable features.
• Construct a concept lattice from the LSI-reduced feature matrix to identify co-occurring and distinct features across diagram variants.
• Integrate FCA and LSI to enhance the precision of commonality and variability detection in use-case diagrams.
• Validate the approach using a mobile media application case study with multiple diagram versions.

Experimental results

Research questions

• RQ1 How can common and variable features be automatically detected across multiple versions of use-case diagrams?
• RQ2 To what extent does the integration of LSI and FCA improve the accuracy of detecting commonality and variability in use-case diagrams?
• RQ3 Can the proposed method effectively support software engineers in understanding feature evolution during product line reengineering?
• RQ4 How does the approach perform when applied to a real-world software case study?

Key findings

• The proposed approach successfully detected all common and variable features in the mobile media case study with high precision.
• Formal concept analysis effectively visualized the relationships between common and variable features, enabling clear interpretation by software engineers.
• Latent semantic indexing enhanced the detection of semantic similarities among use-cases, improving feature grouping accuracy.
• The integration of LSI and FCA significantly improved the robustness and reliability of feature detection across diagram variants.
• The method demonstrated strong performance in identifying both shared and unique features across multiple versions of use-case diagrams.
• The validation on a real-world case study confirmed the practical applicability and effectiveness of the proposed approach.