Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


48235.pdf (1.92 MB)

ETD Abstract Container

Abstract Header

A Hybrid Approach for using Natural Language Processing Techniques to Assess User Feedback on Static Analysis Tools

Yeboah, Jones
http://orcid.org/0009-0001-5516-258X
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1721232177159388

Abstract Details

2024, PhD, University of Cincinnati, Education, Criminal Justice, and Human Services: Information Technology.
In the field of software development, Static Analysis Tools (SAT) have become increasingly popular for identifying and fixing potential issues in code. These tools help improve code quality and reduce the occurrence of bugs and vulnerabilities. However, evaluating the effectiveness of SAT can be challenging due to the subjective nature of user reviews. To address this challenge, we selected four popular SATs as case studies and they include SonarQube, FindBugs, Checkstyle, and PMD. The first part of this research involves conducting an empirical study for evaluating the performance of SATs. We compared the performance of the four SATs in detecting software defects in diverse open-source Java projects. The study results show that SonarQube performs considerably better than all other tools in defect detection. The second part of this research focuses on the user perspective by evaluating the performance of the SATs through sentiment analysis of user reviews. The study found that user sentiment is a valuable indicator of a tool's effectiveness and reliability. Positive user feedback typically corresponds to higher performance ratings, reflecting greater user satisfaction and tool efficiency. Conversely, negative sentiments often point to performance issues and user dissatisfaction. Thus, incorporating sentiment analysis can provide meaningful insights into the perceived quality and performance of SAT. In the third study, we applied topic modeling techniques to user reviews of SATs. Our analysis highlights the key aspects that users find beneficial and areas where improvements are needed. The findings provide valuable insights into user concerns and preferences, informing the development of more user-friendly and effective SAT. In our fourth study, we propose a theoretical framework that integrates sentiment analysis, composite sentiment score, topic modeling, and emotion detection to extract meaningful insights from user feedback. By quantifying polarity and subjectivity, the framework generates a single assessment score for each review, allowing for segmentation into positive and negative sentiment groups. Through topic modeling, prevalent themes in user feedback are identified, providing valuable insights into user perceptions of the tools. We then validated the effectiveness of the theoretical framework proposed in this research by developing a review engine to test how effectively it evaluates user reviews of SATs. This review engine extracts meaningful insights from user feedback, enhancing the development of future static analysis tools.
Saheed Popoola, Ph.D. (Committee Chair)
Yanran Liu, Ph.D. (Committee Member)
Isaac Kofi Nti, Ph.D. (Committee Member)
M. Murat Ozer, Ph.D. (Committee Member)
152 p.
Information Technology
topic modeling; sentiment analysis; emotion detection; static code analysis; composite sentiment score; user reviews

Recommended Citations

Citations

  • Yeboah, J. (2024). A Hybrid Approach for using Natural Language Processing Techniques to Assess User Feedback on Static Analysis Tools [Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1721232177159388

    APA Style (7th edition)

  • Yeboah, Jones. A Hybrid Approach for using Natural Language Processing Techniques to Assess User Feedback on Static Analysis Tools. 2024. University of Cincinnati, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1721232177159388.

    MLA Style (8th edition)

  • Yeboah, Jones. "A Hybrid Approach for using Natural Language Processing Techniques to Assess User Feedback on Static Analysis Tools." Doctoral dissertation, University of Cincinnati, 2024. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1721232177159388

    Chicago Manual of Style (17th edition)

ucin1721232177159388
61
© 2024, some rights reserved.Creative Commons License A Hybrid Approach for using Natural Language Processing Techniques to Assess User Feedback on Static Analysis Tools by Jones Yeboah is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by University of Cincinnati and OhioLINK.