Feature Location in a Collection of Product Variants: Combining Information Retrieval and Hierarchical Clustering

Hamzeh Eyal-Salman 1 Abdelhak-Djamel Seriai 1 Christophe Dony 1
1 MAREL - Models And Reuse Engineering, Languages
LIRMM - Laboratoire d'Informatique de Robotique et de Microélectronique de Montpellier
Abstract : Locating source code elements relevant to a given fea- ture is an important step in the process of re-engineering software variants, developed by an ad-hoc reuse technique, into a Software Product Line (SPL) for systematic reuse. Existing works on using Information Retrieval (IR) tech- niques do not consider the abstraction gap between feature and source code levels. In our recent work, we have im- proved the effectiveness of IR-based feature location by in- troducing an intermediate level between feature and source code levels, called “code-topics”. We used Formal Con- cept Analysis (FCA) to identify such “code-topics”. In this paper, we investigate the results of using Agglomerative Hierarchical Clustering (AHC) algorithm to identify code- topics. In our experimental evaluation, we show that AHC significantly increases the recall of feature location with a minor decrease of precision compared to FCA.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [10 references]  Display  Hide  Download

https://hal-lirmm.ccsd.cnrs.fr/lirmm-01291261
Contributor : Abdelhak-Djamel Seriai <>
Submitted on : Monday, March 21, 2016 - 11:20:40 AM
Last modification on : Friday, November 23, 2018 - 12:52:07 PM
Document(s) archivé(s) le : Wednesday, June 22, 2016 - 10:40:52 AM

File

seke14paper_232.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : lirmm-01291261, version 1

Collections

Citation

Hamzeh Eyal-Salman, Abdelhak-Djamel Seriai, Christophe Dony. Feature Location in a Collection of Product Variants: Combining Information Retrieval and Hierarchical Clustering. SEKE: Software Engineering and Knowledge Engineering, Jul 2014, Vancouver, Canada. pp.426-430. ⟨lirmm-01291261⟩

Share

Metrics

Record views

157

Files downloads

256