(Automated) software modularization using community detection

Klaus Marius Hansen; Konstantinos Manikas

doi:10.1007/978-3-319-23727-5_8

(Automated) software modularization using community detection

Klaus Marius Hansen, Konstantinos Manikas

Datalogisk Institut

1 Citationer (Scopus)

Abstract

The modularity of a software system is known to have an effect on, among other, development effort, change impact, and technical debt. Modularizing a specific system and evaluating this modularization is, however, challenging. In this paper, we apply community detection methods to the graph of class dependencies in software systems to find optimal modularizations through communities. We evaluate this approach through a study of 111 Java systems contained in the Qualitas Corpus. We found that using the modularity function of Newman with an Erdős-Rényi null-model and using the community detection algorithm of Reichardt and Bornholdt improved community quality for all systems, that coupling decreased for 99 of the systems, and that coherence increased for 102 of the systems. Furthermore, the modularity function correlates with existing metrics for coupling and coherence.

Originalsprog	Engelsk
Titel	Software architecture : 9th European Conference, ECSA 2015, Dubrovnik/Cavtat, Croatia, September 7–11, 2015, Proceedings
Redaktører	Danny Weyns, Raffaela Mirandola, Ivica Crnkovic
Antal sider	8
Forlag	Springer
Publikationsdato	2015
Sider	95-102
Kapitel	8
ISBN (Trykt)	978-3-319-23726-8
ISBN (Elektronisk)	978-3-319-23727-5
DOI	https://doi.org/10.1007/978-3-319-23727-5_8
Status	Udgivet - 2015
Begivenhed	European Conference on Software Architecture Workshops 2015 - Dubrovnik/Cavtat, Kroatien Varighed: 7 sep. 2015 → 11 sep. 2015 Konferencens nummer: 9

Konference

Konference	European Conference on Software Architecture Workshops 2015
Nummer	9
Land/Område	Kroatien
By	Dubrovnik/Cavtat
Periode	07/09/2015 → 11/09/2015

Navn	Lecture notes in computer science
Vol/bind	9278
ISSN	0302-9743

Adgang til dokumentet

10.1007/978-3-319-23727-5_8

Citationsformater

Hansen, K. M., & Manikas, K. (2015). (Automated) software modularization using community detection. I D. Weyns, R. Mirandola, & I. Crnkovic (red.), Software architecture: 9th European Conference, ECSA 2015, Dubrovnik/Cavtat, Croatia, September 7–11, 2015, Proceedings (s. 95-102). Springer. Lecture notes in computer science Bind 9278 https://doi.org/10.1007/978-3-319-23727-5_8

(Automated) software modularization using community detection. / Hansen, Klaus Marius; Manikas, Konstantinos.

Software architecture: 9th European Conference, ECSA 2015, Dubrovnik/Cavtat, Croatia, September 7–11, 2015, Proceedings. red. / Danny Weyns; Raffaela Mirandola; Ivica Crnkovic. Springer, 2015. s. 95-102 (Lecture notes in computer science, Bind 9278).

Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › peer review

Hansen, KM & Manikas, K 2015, (Automated) software modularization using community detection. i D Weyns, R Mirandola & I Crnkovic (red), Software architecture: 9th European Conference, ECSA 2015, Dubrovnik/Cavtat, Croatia, September 7–11, 2015, Proceedings. Springer, Lecture notes in computer science, bind 9278, s. 95-102, European Conference on Software Architecture Workshops 2015, Dubrovnik/Cavtat, Kroatien, 07/09/2015. https://doi.org/10.1007/978-3-319-23727-5_8

@inproceedings{a9fe23592823483db36bba478c7fbf2b,

title = "(Automated) software modularization using community detection",

abstract = "The modularity of a software system is known to have an effect on, among other, development effort, change impact, and technical debt. Modularizing a specific system and evaluating this modularization is, however, challenging. In this paper, we apply community detection methods to the graph of class dependencies in software systems to find optimal modularizations through communities. We evaluate this approach through a study of 111 Java systems contained in the Qualitas Corpus. We found that using the modularity function of Newman with an Erd{\H o}s-R{\'e}nyi null-model and using the community detection algorithm of Reichardt and Bornholdt improved community quality for all systems, that coupling decreased for 99 of the systems, and that coherence increased for 102 of the systems. Furthermore, the modularity function correlates with existing metrics for coupling and coherence.",

author = "Hansen, {Klaus Marius} and Konstantinos Manikas",

year = "2015",

doi = "10.1007/978-3-319-23727-5_8",

language = "English",

isbn = "978-3-319-23726-8",

series = "Lecture notes in computer science",

publisher = "Springer",

pages = "95--102",

editor = "Danny Weyns and Raffaela Mirandola and Ivica Crnkovic",

booktitle = "Software architecture",

note = "European Conference on Software Architecture Workshops 2015 ; Conference date: 07-09-2015 Through 11-09-2015",

}

TY - GEN

T1 - (Automated) software modularization using community detection

AU - Hansen, Klaus Marius

AU - Manikas, Konstantinos

N1 - Conference code: 9

PY - 2015

Y1 - 2015

N2 - The modularity of a software system is known to have an effect on, among other, development effort, change impact, and technical debt. Modularizing a specific system and evaluating this modularization is, however, challenging. In this paper, we apply community detection methods to the graph of class dependencies in software systems to find optimal modularizations through communities. We evaluate this approach through a study of 111 Java systems contained in the Qualitas Corpus. We found that using the modularity function of Newman with an Erdős-Rényi null-model and using the community detection algorithm of Reichardt and Bornholdt improved community quality for all systems, that coupling decreased for 99 of the systems, and that coherence increased for 102 of the systems. Furthermore, the modularity function correlates with existing metrics for coupling and coherence.

AB - The modularity of a software system is known to have an effect on, among other, development effort, change impact, and technical debt. Modularizing a specific system and evaluating this modularization is, however, challenging. In this paper, we apply community detection methods to the graph of class dependencies in software systems to find optimal modularizations through communities. We evaluate this approach through a study of 111 Java systems contained in the Qualitas Corpus. We found that using the modularity function of Newman with an Erdős-Rényi null-model and using the community detection algorithm of Reichardt and Bornholdt improved community quality for all systems, that coupling decreased for 99 of the systems, and that coherence increased for 102 of the systems. Furthermore, the modularity function correlates with existing metrics for coupling and coherence.

U2 - 10.1007/978-3-319-23727-5_8

DO - 10.1007/978-3-319-23727-5_8

M3 - Article in proceedings

SN - 978-3-319-23726-8

T3 - Lecture notes in computer science

SP - 95

EP - 102

BT - Software architecture

A2 - Weyns, Danny

A2 - Mirandola, Raffaela

A2 - Crnkovic, Ivica

PB - Springer

T2 - European Conference on Software Architecture Workshops 2015

Y2 - 7 September 2015 through 11 September 2015

ER -