A Contextualized Vocabulary Model for Identifying Technical Debt in - - PowerPoint PPT Presentation

A Contextualized Vocabulary Model for Identifying Technical Debt in Code Comments.

Mário André, André Batista, Manoel Mendonça, Rodrigo Spínola

Federal University of Bahia

Agenda.

• Introduction

– Motivation

• A Contextualized Vocabulary Model for

Identifying Technical Debt (CVM-TD)

• Exploratory Study
• Conclusion and Future Works

www.tdresearchteam.com 2

Motivation.

• Different indicators have been used by

automated approaches to identify TD[2];

–These indicators use software metrics.

• The problem with these approaches...
• In order to complement these quantitative

analyses, we developed the CVM-TD

www.tdresearchteam.com 3

Agenda.

• Introduction

– Motivation

• A Contextualized Vocabulary Model for Identifying

Technical Debt (CVM-TD)

• Exploratory Study
• Conclusion and Future Works

www.tdresearchteam.com 4

CVM-TD

• CVM-TD is a contextualized collection of terms

that focuses on using word classes and code tags.

www.tdresearchteam.com 5

CVM-TD

www.tdresearchteam.com 6

To generate a NP, we combine [Verb] + [Adj ] + [SE Noun] SE noun is the central element in a NP.

e.g. bad (Adj) + code (Noun). remove (Verb) + method (Noun).

• NP is a composed terms that provide

high context information

Noun Adj Verb

CVM-TD

www.tdresearchteam.com 7

We combine: Expression = [Adv] + [Tag] + [NP] Expression = [Adv] + [Tag] + SE Noun

e.g.: “Todo: should fix the bad code” Here, we have a contextualized expression.

CVM-TD.

• CVM-TD provides a set of TD vocabulary that may

be used to filter comments that need more attention because they may indicate a TD.

• The contextualized vocabulary can be used to

support TD identification

www.tdresearchteam.com 8

Agenda.

• Introduction

– Motivation

• A Contextualized Vocabulary Model for

Identifying Technical Debt (CVM-TD)

• Exploratory Study
• Conclusion and Future Works

www.tdresearchteam.com 9

Exploratory Study

• We performed an exploratory study to analyze

the viability of the proposed model to support comments analysis for TD identification

Exploratory Study.

Selecting the projects and participants Extracting comments Filtering comments Analyzing comments

• jEdit and

Lucene;

• Four masters

students analyzed the projects;

Set of comments Set of candidate comments Set of comments that can indicate a TD item

Agenda.

• Introduction

– Motivation

• A Contextualized Vocabulary Model for

Identifying Technical Debt (CVM-TD)

• Exploratory Study
• Conclusion and Future Works

www.tdresearchteam.com 12

Contributions of work

• Our findings show that:

– (i) the dimensions considered by the model are used by developers when writing comments, – (ii) CVM-TD provides a vocabulary which may be used to detect TD, and – (iii) In spite of our outcomes being promising, the proposed model needs to be improved and evaluated with other projects.

www.tdresearchteam.com 13

Future works.

• Identify points that may improve CVM-TD;
• Implement CVM-TD model and integrate it with a

comment analysis tool;

• Evaluate the use of CVM-TD in other software

projects.

www.tdresearchteam.com 14

References

[1] C. Izurieta, A. Vetrò, N. Zazworka, Y. Cai, C. Seaman, and F. Shull, "Organizing the technical debt landscape", 3rd Int. Work. Manag. Tech. Debt, MTD 2012 - Proc., pp. 23–26, 2012. [2] N. S. R. Alves, L. F. Ribeiro, V. Caires, T. S. Mendes, and R. O. Spínola, "Towards an Ontology of Terms on Technical Debt". In Proceedings of the Sixth International Workshop on Managing Technical Debt (MTD). IEEE Computer Society, Washington, DC, 2014. [3] P. Wongthongtham, E. Chang, T. Dillon, and I. Sommerville, “Development of a software engineering

• ntology for multisite software development,” IEEE Trans. Knowl. Data Eng., vol. 21, no. 8, pp. 1205–1217,

2009. [4] M. Colaço, M. Mendonça, M. A. F. Farias, and P. Henrique, “A Neurolinguistic-based Methodology for Identifying OSS Developers Context-Specific Preferred Representational Systems,” ICSEA, 2012

www.tdresearchteam.com 15

Questions

www.tdresearchteam.com 16

A Contextualized Vocabulary Model for Identifying Technical Debt on Code Comments

Mário André, André Batista, Manoel Mendonça, Rodrigo Spínola

Federal University of Bahia