Academia.eduAcademia.edu

Figure 5 – uploaded by Carolyn Seaman

See full PDFdownloadDownload figure
5.2.1 Evaluation Studies

Figure 5 5.2.1 Evaluation Studies

Related Figures (20)

Figure 1. The systematic mapping process (adapted from [Petersen et al., 2008]).
Figure 1. The systematic mapping process (adapted from [Petersen et al., 2008]).
Table 2. Generic search String. 4.3. Data Sources For these keywords, the search string shown in Table 2 was defined. We applied this search string to Titles and Abstracts. We chose not to do full text search because we found that full text search resulted in a very large number of studies from domains other than software development.
Table 2. Generic search String. 4.3. Data Sources For these keywords, the search string shown in Table 2 was defined. We applied this search string to Titles and Abstracts. We chose not to do full text search because we found that full text search resulted in a very large number of studies from domains other than software development.
4.5 Classification scheme
4.5 Classification scheme
Table 4. Indicators organized by TD type
Table 4. Indicators organized by TD type
increased confidence. Among the indicators evaluated, we can highlight god class, reported by six studies (three controlled experiments and three case studies). This suggests that the potential usefulness of god class in supporting the activities of TD identification in software projects is already established. Another indicator that stands out is ASA issues, which were evaluated through five case studies and two controlled experiments. Furthermore, Documentation and Software Architecture Issues have been evaluated in five case studies. Violation of Modularity and Structural Dependency indicators have been used in four case studies.
increased confidence. Among the indicators evaluated, we can highlight god class, reported by six studies (three controlled experiments and three case studies). This suggests that the potential usefulness of god class in supporting the activities of TD identification in software projects is already established. Another indicator that stands out is ASA issues, which were evaluated through five case studies and two controlled experiments. Furthermore, Documentation and Software Architecture Issues have been evaluated in five case studies. Violation of Modularity and Structural Dependency indicators have been used in four case studies.
Figure 9. Papers by identified data source
Figure 9. Papers by identified data source
The number of published papers per year discussing management strategies was also investigated. In 2010, no management strategy was identified in any published paper. Only a handful of TD management papers were published in 2011, introducing such ideas as cost-benefit analysis and portfolio management. In 2012, the number of papers, and the number of newly introduced management techniques, exploded, with 16 papers published that year. In 2013, a total of 10 TD management papers were identified. Finally, in 2014, the number of papers exploring TD management was also high reaching 17 papers.
The number of published papers per year discussing management strategies was also investigated. In 2010, no management strategy was identified in any published paper. Only a handful of TD management papers were published in 2011, introducing such ideas as cost-benefit analysis and portfolio management. In 2012, the number of papers, and the number of newly introduced management techniques, exploded, with 16 papers published that year. In 2013, a total of 10 TD management papers were identified. Finally, in 2014, the number of papers exploring TD management was also high reaching 17 papers.
Figure 10. Papers by management strategy and type of study. Figure 10 shows the number of papers on management strategies and the type of performed evaluation. It can be observed that the studies are concentrated on case study method. Besides, not all strategies were evaluated. This implies that most of the proposals in the TD management area still require more experiments, so that their benefits and limitations can be known with increased confidence. 5.3.2 Software visualization techniques
Figure 10. Papers by management strategy and type of study. Figure 10 shows the number of papers on management strategies and the type of performed evaluation. It can be observed that the studies are concentrated on case study method. Besides, not all strategies were evaluated. This implies that most of the proposals in the TD management area still require more experiments, so that their benefits and limitations can be known with increased confidence. 5.3.2 Software visualization techniques
Figure 11. Types of software visualization techniques proposed to manage TD. As can be seen in Figure 11, 22 of the 100 primary studies proposed software visualization techniques in the context of TD management. The most proposed visualization techniques were dependency matrix, bar graph, and pie chart format. One opportunity that arises here is to investigate different types of software visualization techniques already proposed in other contexts in software maintenance and evolution [Novais et al. 2013], and adapt them to manage TD.
Figure 11. Types of software visualization techniques proposed to manage TD. As can be seen in Figure 11, 22 of the 100 primary studies proposed software visualization techniques in the context of TD management. The most proposed visualization techniques were dependency matrix, bar graph, and pie chart format. One opportunity that arises here is to investigate different types of software visualization techniques already proposed in other contexts in software maintenance and evolution [Novais et al. 2013], and adapt them to manage TD.
Table 7. Intersection among the studies for Research Question 3 (TD Management)
Table 7. Intersection among the studies for Research Question 3 (TD Management)
Regarding “(Q2.1 and Q3.1) Which empirical evaluations have been performed?” there are no similar research questions in the other studies. However, we analyzed each of the other studies more closely in order to identify possible intersections. Li et al. (2015) was the only study that took into consideration if the identified technique had been evaluated. The authors used this information to represent a general view of the evidence levels of their selected studies. In contrast, in our work we used this information to characterize the level of confidence we have about TD indicators and TD management strategies.
Regarding “(Q2.1 and Q3.1) Which empirical evaluations have been performed?” there are no similar research questions in the other studies. However, we analyzed each of the other studies more closely in order to identify possible intersections. Li et al. (2015) was the only study that took into consideration if the identified technique had been evaluated. The authors used this information to represent a general view of the evidence levels of their selected studies. In contrast, in our work we used this information to characterize the level of confidence we have about TD indicators and TD management strategies.
Related topics:
Information SystemsINFORMATION TECHNOLOGY AND SOFTWARE DEVELOPMENT
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved