LLM Contribution Summarization in Software Projects

Diego Pavan Soler; Fabio Roberto de Miranda; Rafael Corsi Ferrao

REVIEW 3 major objections 5 minor 32 references

LLM Contribution Summarization in Software Projects

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A fully automated pipeline can summarize each student's code contributions during intensive team projects, and instructors found the results usable for monitoring.

desk verdict An honest, cheap, well-documented experience report on LLM-based commit summarization; the tool is plausible and useful for instructors, but the accuracy claim is weaker than the paper's conclusion. read the letter →

arxiv 2505.17710 v1 pith:QEIXFCZJ submitted 2025-05-23 cs.SE cs.CY

Rafael Corsi Ferrao , Fabio Roberto de Miranda , Diego Pavan Soler This is my paper · ORCID

classification cs.SEcs.CY

keywords LLMsummarizationcontributionanalysisproject-basedlearningsoftwareengineeringeducationGitrepositoryminingautomatedfeedbackteamworkassessmentlargelanguagemodels

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that an automated pipeline can translate version-control history into readable, per-student summaries of code work, giving instructors a timely view of open-ended team projects when a uniform code structure cannot be imposed. The pipeline was run in two three-week sprints with real external clients, producing 130 weekly summaries for 65 students at a total API cost of roughly $4 per semester. The authors report that the summaries captured a large proportion of the activities students actually performed, with eight partial and four total omissions and one misattribution; students and instructors both recognized the reports as largely consistent with their own knowledge. They also report that instructors considered the summaries useful for keeping up with teams, though the accuracy numbers rest on student self-reviews whose thoroughness the authors themselves flag as uncertain.

What carries the argument

The load-bearing mechanism is a two-level compression of Git blame data before any final summary is written. A first, lighter model reads each file and emits a functionality table (purpose, difficulty, size, complexity); a second model reads who changed what within each file and emits a contribution table; a stronger model then synthesizes individual and team summaries from those tables, using predefined role classifications only optionally. This preprocessing exists because feeding raw Git blame output directly to the LLM caused hallucinations and information loss. The tables are the intermediate representation that lets the final model summarize without reading raw line-level history.

What would settle it

An independent audit of the same 130 summaries—with a researcher reconstructing each student's actual contributions from commits, diffs, tasks, and merge history, and comparing that reconstruction to the LLM output—would establish whether the reported omission counts are accurate.

Watch

Extended reading notes

Core claim

The central claim is that the combination of Git-blame-based contribution extraction and a two-stage LLM summarization chain is a workable mechanism for continuous, objective monitoring of individual work in project courses. In the authors' evaluation, the tool answered the first research question 'partially yes': it reports, in large proportion, activities actually performed, while failing some contributions; it answered the second 'yes': instructors found the summaries a useful potential tool to keep up with the projects. The paper also surfaces a failure mode they call 'comment injections,' in which a code comment, rather than the actual commit author, led the LLM to attribute a feature to the wrong student.

Load-bearing premise

The accuracy conclusion depends on the assumption that students, during the 120-minute review sessions, actually detected and reported most errors in their own summaries, even though the authors note the thoroughness of those reviews was uncertain.

Editorial extensions

If this is right

Instructors can monitor many teams at once during intensive sprints, because the summaries compress line-level history into short narratives focused on one student at a time.
Because the full semester's summaries cost about $4, running the pipeline weekly—or even more often—is financially viable for routine formative feedback.
Students can use the externally generated summary as neutral evidence in team discussions, as illustrated by one group confronting a teammate whose contribution was judged minimal.
The 'comment injection' case implies that any automation basing attribution on file content must treat comments as unverified claims; otherwise authorship can be assigned to the writer of a comment rather than the committer.
Minor tasks and automated scaffold code being overvalued is a known bias of the current prompt structure, and correcting it is part of making the tool fairer.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural follow-up is to measure recall and precision against a task-level ground truth built from the teams' Scrum boards, rather than from student self-reports; this would tell whether eight partial and four total omissions are the true error counts.
The same two-stage compression could be adapted to non-code contributions, such as documentation, issue discussion, or review comments, provided each artifact can be attributed to a person.
If the 'comment injection' failure generalizes, a simple countermeasure is to derive attribution from the diff itself and ignore comment text when assigning authorship; the paper describes the failure but does not propose this fix.
The cost figure opens the door to running summaries in near-real time; the authors state real-time analysis is the long-term goal, and the $4-per-semester figure suggests the infrastructure, not the API bill, is the binding constraint.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The one thing you should know: this is a straightforward experience report, not a rigorous efficacy study. What is actually new is the full-scale deployment—65 students, two semesters, 130 summaries, total cost $4—and the preprocessing insight that feeding raw git blame data to an LLM causes hallucinations. The pipeline description is clear and reproducible enough that someone could rebuild it. The authors are also commendably honest about limitations: they explicitly flag the uncertainty in students' self-review thoroughness (Section VI) and the attribution problems from pair programming and squashed commits (Section V-A). That kind of candor is worth crediting.

The soft spot is exactly where the stress-test lands. The central claim in the abstract and Section VII—that the tool "reports, in large proportion," activities actually performed—is not supported by the quantitative evidence. The support is 8 partial and 4 total omissions reported by students during a group feedback session, with the Scrum master relaying issues. That is a summary-level count, not a recall measure. A summary with no reported omission could still omit most of a student's work. The 8/130 number does not mean 94% of activities were captured; it means 94% of summaries had no reported omission, which is a different claim. The self-selected interview sample (8 of 65) and the instructors' impressions are directionally positive but cannot carry the accuracy estimate either. So the paper's own conditional answer to Q1—"partially yes"—is the right level of confidence, but the phrasing "in large proportion" goes further than the data allows.

That said, the paper never over-engineers its claims. It is a pilot, it says so, and the qualitative evidence (students feeling recognized, teams using summaries to have difficult conversations) is plausible and useful. The cost and simplicity make it a sensible tool for an instructor to try.

Who gets value: software engineering education researchers and instructors running project courses. It is not a methods paper and doesn't need to be. The citation pattern is appropriate—the authors cite prior repo-mining and LLM-in-education work, including their own SIGCSE Virtual 2024 short paper, which is the right thing to do.

My recommendation: send it to review. An editor should not desk-reject this; it is a competent experience report with honest limitations and a clear practical contribution. The reviewers should ask the authors to soften the "large proportion" language and add explicit denominators, but the paper deserves referee time.

Referee Report

3 major / 5 minor

Summary. The paper presents an LLM-based tool that summarizes individual student code contributions in team software projects by mining Git repositories with PyDriller, preprocessing blame data with GPT-4o-mini, and generating final summaries with GPT-4o. The tool was deployed in two semesters of a 3-week sprint (65 students, 130 weekly summaries). The authors report 8 partial omissions, 4 total omissions, and 1 factual inaccuracy across these summaries, and they use student and instructor feedback to argue that the tool is useful and reports activities in large proportion. The paper also discusses limitations, including uncertain review thoroughness, role-classification confusion, and inflation of minor tasks.

Significance. If the accuracy claim held, this would be a useful, low-cost ($4 per semester) innovation for project-based courses, combining repository mining with LLM summarization to give instructors and students regular feedback. The paper's strengths are its detailed pipeline description, the explicit acknowledgment of failure modes (comment injection, pair-programming attribution, role inflation), and the identification of concrete social benefits such as facilitating fair task-distribution discussions. However, the central quantitative claim is not currently supported by the evidence: the error counts come from self-selected, self-reported student review without a denominator, and no baseline is provided. The contribution is therefore best framed as an experience report with qualitative insights, not as a validated accuracy measurement.

major comments (3)

[Section V and VII] The central claim 'The tool reports, in large proportion, activities that were in fact performed by the student' is not supported by the reported data. The counts (8 partial omissions, 4 total omissions, and 1 factual inaccuracy among 130 summaries) are summary-level and lack a denominator of activities; a summary with no reported omission can still capture only a fraction of the student's work. Since Section VI concedes that 'the thoroughness of the students' reviews is uncertain,' the counts are at best lower bounds, so the 'large proportion' claim is not quantitatively established.
[Section V-A] The accuracy evidence rests on 8 self-selected volunteers out of 65 students. This sample is not representative, and the statement in Section V that 'there was general agreement among the students' overreaches the data. The authors should either present the student reactions as anecdotal or obtain a systematic sample (e.g., all students) to estimate accuracy.
[Section V-A and VI] The attribution method (Git blame via PyDriller) fails for pair-programmed or squashed work, as the authors note when discussing co-authorship tags. This known mechanism creates omissions that students may not flag during the review session, so the reported omission counts likely underestimate the true error rate. The paper would need an independent ground truth (e.g., instructor logs, commit-level cross-check) to calibrate the 'in large proportion' conclusion.

minor comments (5)

[Abstract] The abstract says 'failure to detect students contribution'; it should be 'students' contributions'.
[Section V] The phrase 'general agreement among the students' is too strong for the 8-student volunteer sample; consider rewording to 'the interviewed students agreed'.
[Table II] Table II lacks clear column headers; the table mixes student IDs, filenames, and narrative descriptions, making it harder to parse than necessary.
[References] Reference [29] is the authors' own earlier SIGCSE Virtual 2024 paper; the present submission should state explicitly what it adds beyond that publication (e.g., full evaluation, second semester, role classification).
[Section VI] The error data are said to be collected during '120-minute sessions,' but Section V describes two presentations per semester; please clarify whether the results aggregate both weeks and both semesters and how the sessions map to the reported counts.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: accuracy claims rest on self-reported review whose thoroughness the authors concede is uncertain, which is a measurement-validity threat, not a self-referential input.

full rationale

Score 0. The paper's central claim, that the tool reports in large proportion activities actually performed by the student, is an empirical evaluation claim rather than a derived or fitted result. The quantitative support is a count of 8 partial omissions and 4 total omissions across 130 summaries, collected when students reviewed their own summaries (Sections V and VI), with the authors explicitly noting that 'the thoroughness of the students' reviews is uncertain' (Section VI). That is a threat to internal validity, but the ground truth used for comparison is external to the tool's pipeline: student memory and non-author instructor impressions. No parameter is fitted to the summaries and later renamed a prediction; the LLM summaries are produced from git blame data and compared against human review. The self-citations [26], [27], and [29] are contextual or describe the tool's earlier form; none is load-bearing for the accuracy conclusion, and none is used to forbid alternative approaches. The accuracy conclusion therefore has independent empirical content, and the paper's weaknesses are measurement-validity and generalization limitations, not circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

This pipeline has no fitted numeric parameters; instead, it inherits several unstated domain assumptions. The most consequential are that Git blame is a complete and accurate record of contributions, that students self-reviewing their own summaries provide a valid accuracy benchmark, and that Python-based sprints at a single institution are representative of project-based courses. These assumptions, not free parameters, carry the load.

assumptions (3)

domain assumption Git history and blame attribution faithfully capture individual student contributions
The pipeline relies on PyDriller blame output as ground truth for who wrote which line. Co-authored or pair-programmed lines may be misattributed, as acknowledged by student requests for co-authorship tags (Section V-A).
domain assumption Students' self-reports during review sessions are a reliable measure of summary accuracy
Section VI states the thoroughness of students' reviews is uncertain. Error counts (8 partial, 4 total omissions) therefore may undercount, making the claimed accuracy unverified.
domain assumption The sample of 65 students across two semesters generalizes to other project-based courses
All projects were Python-based with frontend/backend/data components; Section VI notes this limits external validity.

how reviews work

0 comments

Cite this review

Pith. "Pith review of LLM Contribution Summarization in Software Projects." pith.science (2026). https://pith.science/paper/QEIXFCZJ

@misc{pith2026250517710,
  author       = {Pith},
  title        = {Pith review of: LLM Contribution Summarization in Software Projects},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QEIXFCZJ}},
  note         = {Machine review of arXiv:2505.17710}
}

read the original abstract

This full paper in innovative practice provides an automated tool to summarize individual code contributions in project-based courses with external clients. Real industry projects offer valuable learning opportunities by immersing students in authentic problems defined by external clients. However, the open-ended and highly variable scope of these projects makes it challenging for instructors and teaching assistants to provide timely and detailed feedback. This paper addresses the need for an automated and objective approach to evaluate individual contributions within team projects. In this paper, we present a tool that leverages a large language model (LLM) to automatically summarize code contributions extracted from version control repositories. The tool preprocesses and structures repository data, and uses PyDriller to isolate individual contributions. Its uniqueness lies in the combination of LLM prompt engineering with automated repository analysis, thus reducing the manual grading burden while providing regular and informative updates. The tool was assessed over two semesters during a three-week, full-time software development sprint involving 65 students. Weekly summaries were provided to teams, and both student and faculty feedback indicated the tool's overall usefulness in informing grading and guidance. The tool reports, in large proportion, activities that were in fact performed by the student, with some failure to detect students' contribution. The summaries were considered by the instructors as a useful potential tool to keep up with the projects.

Figures

Figures reproduced from arXiv: 2505.17710 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 21 canonical work pages

[1]

What Is the Impact of ChatGPT on Education? A Rapid Review of the Literature,

C. K. Lo, “What Is the Impact of ChatGPT on Education? A Rapid Review of the Literature,”Education Sciences, vol. 13, no. 4, p. 410, Apr. 2023, number: 4 Publisher: Multidisciplinary Digital Publishing Institute. [Online]. Available: https://www.mdpi.com/2227-7102/13/4/ 410

work page 2023
[2]

Ways of applying artificial intelligence in software engineering,

R. Feldt, F. G. de Oliveira Neto, and R. Torkar, “Ways of applying artificial intelligence in software engineering,” inProceedings of the 6th International Workshop on Realizing Artificial Intelligence Synergies in Software Engineering, ser. RAISE ’18. New York, NY , USA: Association for Computing Machinery, 2018, pp. 35–41. [Online]. Available: https://d...

work page arXiv 2018
[3]

Generating High-Precision Feedback for Programming Syntax Errors using Large Language Models,

T. Phung, J. Cambronero, S. Gulwani, T. Kohn, R. Majumdar, A. Singla, and G. Soares, “Generating High-Precision Feedback for Programming Syntax Errors using Large Language Models,” Apr. 2023, arXiv:2302.04662 [cs]. [Online]. Available: http://arxiv.org/abs/2302. 04662

arXiv 2023
[4]

ChatGPT for good? On opportunities and challenges of large language models for education,

E. Kasneci, K. Sessler, S. K ¨uchemann, M. Bannert, D. Dementieva, F. Fischer, U. Gasser, G. Groh, S. G ¨unnemann, E. H ¨ullermeier, S. Krusche, G. Kutyniok, T. Michaeli, C. Nerdel, J. Pfeffer, O. Poquet, M. Sailer, A. Schmidt, T. Seidel, M. Stadler, J. Weller, J. Kuhn, and G. Kasneci, “ChatGPT for good? On opportunities and challenges of large language m...

work page 2023
[5]

Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning,

D. Baidoo-Anu and L. Owusu Ansah, “Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning,” Rochester, NY , Jan. 2023. [Online]. Available: https://papers.ssrn.com/abstract= 4337484

work page 2023
[6]

How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment,

A. Gilson, C. W. Safranek, T. Huang, V . Socrates, L. Chi, R. A. Taylor, and D. Chartash, “How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment,”JMIR Medical Education, vol. 9, no. 1, p. e45312, Feb. 2023, company: JMIR Medical Educatio...

work page 2023
[7]

An LLM-based Reflection Analysis Tool for Identifying and Addressing Challenging Topics,

N. Dehbozorgi and M. T. Kunuku, “An LLM-based Reflection Analysis Tool for Identifying and Addressing Challenging Topics,” inProceedings of the 55th ACM Technical Symposium on Computer Science Education V. 2, ser. SIGCSE 2024. New York, NY , USA: Association for Computing Machinery, Mar. 2024, pp. 1618–1619. [Online]. Available: https://dl.acm.org/doi/10....

work page arXiv 2024
[8]

ChatGPT: Bullshit spewer or the end of traditional assessments in higher education?

J. Rudolph, S. Tan, and S. Tan, “ChatGPT: Bullshit spewer or the end of traditional assessments in higher education?”Journal of Applied Learning and Teaching, vol. 6, no. 1, pp. 342–363, Jan. 2023, number:

work page 2023

Show all 32 references

[9]

Available: https://journals.sfu.ca/jalt/index.php/jalt/article/ view/689

[Online]. Available: https://journals.sfu.ca/jalt/index.php/jalt/article/ view/689
[10]

Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies,

B. Cope, M. Kalantzis, and D. Searsmith, “Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies,”Educational Philosophy and Theory, vol. 53, no. 12, pp. 1229–1245, Oct. 2021, publisher: Routledge eprint: https://doi.org/10.1080/00...

2021
[11]

Software Engineering Education Must Adapt and Evolve for an LLM Environment,

V . D. Kirova, C. S. Ku, J. R. Laracy, and T. J. Marlowe, “Software Engineering Education Must Adapt and Evolve for an LLM Environment,” inProceedings of the 55th ACM Technical Symposium on Computer Science Education V. 1. Portland OR USA: ACM, Mar. 2024, pp. 666–672. [Online]...

2024
[12]

Development effort estimation in free/open source software from activity in version control systems,

G. Robles, A. Capiluppi, J. M. Gonzalez-Barahona, B. Lundell, and J. Gamalielsson, “Development effort estimation in free/open source software from activity in version control systems,”Empirical Software Engineering, vol. 27, no. 6, p. 135, Jul. 2022, number: 6. [Online]. Avai...

2022 doi
[13]

Investigating the skill gap between graduating students and industry expectations,

A. Radermacher, G. Walia, and D. Knudson, “Investigating the skill gap between graduating students and industry expectations,” inCompanion Proceedings of the 36th International Conference on Software Engineering, ser. ICSE Companion 2014. New York, NY , USA: Association for Co...

2014
[14]

Communication learning outcomes from software engineering professionals: A basis for teaching communication in the engineering curriculum,

S. Ruf and M. Carter, “Communication learning outcomes from software engineering professionals: A basis for teaching communication in the engineering curriculum,”IEEE, Oct. 2009, accepted: 2010-12- 09T19:44:39Z ISBN: 9781424447152 Publisher: Institute of Electrical and Electro...

2009
[15]

Facilitating Transfer of Skills between Group Projects and Work Teams,

D. R. Ettington and R. R. Camp, “Facilitating Transfer of Skills between Group Projects and Work Teams,”Journal of Management Education, vol. 26, no. 4, pp. 356–379, Aug. 2002, publisher: SAGE Publications Inc. [Online]. Available: https://doi.org/10.1177/105256290202600404

2002 doi
[16]

Lessons from the Best and Worst Student Team Experiences: How a Teacher can make the Difference,

D. R. Bacon, K. A. Stewart, and W. S. Silver, “Lessons from the Best and Worst Student Team Experiences: How a Teacher can make the Difference,”Journal of Management Education, vol. 23, no. 5, pp. 467–488, Oct. 1999, publisher: SAGE Publications Inc. [Online]. Available: https...

1999 doi
[17]

Why Some Groups Fail: a Survey of Students’ Experiences with Learning Groups,

S. B. Feichtner and E. A. Davis, “Why Some Groups Fail: a Survey of Students’ Experiences with Learning Groups,”Organizational Behavior Teaching Review, vol. 9, no. 4, pp. 58–73, Nov. 1984, publisher: SAGE Publications. [Online]. Available: https://journals.sagepub.com/doi/abs...

1984 doi
[18]

Collaborative learning and anxiety: a phenomenographic study of collaborative learning activities,

K. Falkner, N. J. Falkner, and R. Vivian, “Collaborative learning and anxiety: a phenomenographic study of collaborative learning activities,” inProceeding of the 44th ACM technical symposium on Computer science education, ser. SIGCSE ’13. New York, NY , USA: Association for C...

2013
[19]

The Comprehensive Assessment of Team Member Effectiveness: Development of a Behaviorally Anchored Rating Scale for Self- and Peer Evaluation,

M. W. Ohland, M. L. Loughry, D. J. Woehr, L. G. Bullard, R. M. Felder, C. J. Finelli, R. A. Layton, H. R. Pomeranz, and D. G. Schmucker, “The Comprehensive Assessment of Team Member Effectiveness: Development of a Behaviorally Anchored Rating Scale for Self- and Peer Evaluatio...

2012
[20]

“It’s not Fair!

L. E. Willcoxson, ““It’s not Fair!”: Assessing the Dynamics and Resourcing of Teamwork,”Journal of Management Education, vol. 30, no. 6, pp. 798–808, Dec. 2006, publisher: SAGE Publications Inc. [Online]. Available: https://doi.org/10.1177/1052562906287964

2006 doi
[21]

Using Learning Analytics to Visualise Computer Science Teamwork,

H. Tarmazdi, R. Vivian, C. Szabo, K. Falkner, and N. Falkner, “Using Learning Analytics to Visualise Computer Science Teamwork,” in Proceedings of the 2015 ACM Conference on Innovation and Technology in Computer Science Education, ser. ITiCSE ’15. New York, NY , USA: Associati...

2015
[22]

GitCanary: A Tool for Analyzing Student Contributions in Group Programming Assignments,

J. J. Sandee and E. Aivaloglou, “GitCanary: A Tool for Analyzing Student Contributions in Group Programming Assignments,” in Proceedings of the 20th Koli Calling International Conference on Computing Education Research, ser. Koli Calling ’20. New York, NY , USA: Association fo...

2020
[23]

Measuring developer contribution from software repository data,

G. Gousios, E. Kalliamvakou, and D. Spinellis, “Measuring developer contribution from software repository data,” inProceedings of the 2008 international working conference on Mining software repositories. Leipzig Germany: ACM, May 2008, pp. 129–132. [Online]. Available: https:...

2008
[24]

Assessing developer contribution with repository mining-based metrics,

J. Lima, C. Treude, F. F. Filho, and U. Kulesza, “Assessing developer contribution with repository mining-based metrics,” in 2015 IEEE International Conference on Software Maintenance and Evolution (ICSME), Sep. 2015, pp. 536–540. [Online]. Available: https://ieeexplore.ieee.o...

2015
[25]

Assessing Individual Contributions to Software Engineering Projects with Git Logs and User Stories,

K. Buffardi, “Assessing Individual Contributions to Software Engineering Projects with Git Logs and User Stories,” inProceedings of the 51st ACM Technical Symposium on Computer Science Education, ser. SIGCSE ’20. New York, NY , USA: Association for Computing Machinery, 2020, p...

2020
[26]

Evaluating Commit, Issue and Product Quality in Team Software Development Projects,

C. Hundhausen, A. Carter, P. Conrad, A. Tariq, and O. Adesope, “Evaluating Commit, Issue and Product Quality in Team Software Development Projects,” inProceedings of the 52nd ACM Technical Symposium on Computer Science Education, ser. SIGCSE ’21. New York, NY , USA: Associatio...

2021 doi
[27]

Evaluating mastery-oriented grading in an intensive cs1 course,

I. dos Santos Montagner, R. Corsi Ferr ˜ao, A. Kurauchi, M. Silva, and C. Zilles, “Evaluating mastery-oriented grading in an intensive cs1 course,” inProceedings of the 55th ACM Technical Symposium on Computer Science Education V . 1, ser. SIGCSE 2024. New York, NY , USA: Asso...

2024
[28]

Developing Teamwork With the Aid of Reflection,

G. Cliquet, F. Miranda, G. Tonin, and G. Rodrigues, “Developing Teamwork With the Aid of Reflection,” S ˜ao Paulo,
[29]

Llm- based individual contribution summarization in software projects,

F. de Miranda, R. C. Ferrao, D. P. Soler, and M. A. Vieira Graglia, “Llm- based individual contribution summarization in software projects,” in Proceedings of the 2024 on ACM Virtual Global Computing Education Conference V . 2, ser. SIGCSE Virtual 2024. New York, NY , USA: Ass...

2024
[30]

Effects of an online team project-based learning environment with group awareness and peer evaluation on socially shared regulation of learning and self-regulated learning,

J.-W. Lin, “Effects of an online team project-based learning environment with group awareness and peer evaluation on socially shared regulation of learning and self-regulated learning,”Behaviour & Information Technology, vol. 37, no. 5, pp. 445–461, May 2018, publisher: Taylor...

2018
[32]

Pydriller: Python framework for mining software repositories,

D. Spadini, M. Aniche, and A. Bacchelli, “Pydriller: Python framework for mining software repositories,” inProceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, ser. ESEC/FSE 2018. N...

2018
[2023]

Available: https://www.insper.edu.br/en/paee-ale-2023/ proceedings/PAEE ALE 2023 PROCEEDINGS pre print.pdf

[Online]. Available: https://www.insper.edu.br/en/paee-ale-2023/ proceedings/PAEE ALE 2023 PROCEEDINGS pre print.pdf

2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

What Is the Impact of ChatGPT on Education? A Rapid Review of the Literature,

C. K. Lo, “What Is the Impact of ChatGPT on Education? A Rapid Review of the Literature,”Education Sciences, vol. 13, no. 4, p. 410, Apr. 2023, number: 4 Publisher: Multidisciplinary Digital Publishing Institute. [Online]. Available: https://www.mdpi.com/2227-7102/13/4/ 410

work page 2023

[2] [2]

Ways of applying artificial intelligence in software engineering,

R. Feldt, F. G. de Oliveira Neto, and R. Torkar, “Ways of applying artificial intelligence in software engineering,” inProceedings of the 6th International Workshop on Realizing Artificial Intelligence Synergies in Software Engineering, ser. RAISE ’18. New York, NY , USA: Association for Computing Machinery, 2018, pp. 35–41. [Online]. Available: https://d...

work page arXiv 2018

[3] [3]

Generating High-Precision Feedback for Programming Syntax Errors using Large Language Models,

T. Phung, J. Cambronero, S. Gulwani, T. Kohn, R. Majumdar, A. Singla, and G. Soares, “Generating High-Precision Feedback for Programming Syntax Errors using Large Language Models,” Apr. 2023, arXiv:2302.04662 [cs]. [Online]. Available: http://arxiv.org/abs/2302. 04662

arXiv 2023

[4] [4]

ChatGPT for good? On opportunities and challenges of large language models for education,

E. Kasneci, K. Sessler, S. K ¨uchemann, M. Bannert, D. Dementieva, F. Fischer, U. Gasser, G. Groh, S. G ¨unnemann, E. H ¨ullermeier, S. Krusche, G. Kutyniok, T. Michaeli, C. Nerdel, J. Pfeffer, O. Poquet, M. Sailer, A. Schmidt, T. Seidel, M. Stadler, J. Weller, J. Kuhn, and G. Kasneci, “ChatGPT for good? On opportunities and challenges of large language m...

work page 2023

[5] [5]

Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning,

D. Baidoo-Anu and L. Owusu Ansah, “Education in the Era of Generative Artificial Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching and Learning,” Rochester, NY , Jan. 2023. [Online]. Available: https://papers.ssrn.com/abstract= 4337484

work page 2023

[6] [6]

How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment,

A. Gilson, C. W. Safranek, T. Huang, V . Socrates, L. Chi, R. A. Taylor, and D. Chartash, “How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment,”JMIR Medical Education, vol. 9, no. 1, p. e45312, Feb. 2023, company: JMIR Medical Educatio...

work page 2023

[7] [7]

An LLM-based Reflection Analysis Tool for Identifying and Addressing Challenging Topics,

N. Dehbozorgi and M. T. Kunuku, “An LLM-based Reflection Analysis Tool for Identifying and Addressing Challenging Topics,” inProceedings of the 55th ACM Technical Symposium on Computer Science Education V. 2, ser. SIGCSE 2024. New York, NY , USA: Association for Computing Machinery, Mar. 2024, pp. 1618–1619. [Online]. Available: https://dl.acm.org/doi/10....

work page arXiv 2024

[8] [8]

ChatGPT: Bullshit spewer or the end of traditional assessments in higher education?

J. Rudolph, S. Tan, and S. Tan, “ChatGPT: Bullshit spewer or the end of traditional assessments in higher education?”Journal of Applied Learning and Teaching, vol. 6, no. 1, pp. 342–363, Jan. 2023, number:

work page 2023