Pith. sign in

REVIEW 4 major objections 4 minor 96 references

Understanding Developer Pain Points in Federated Learning: Insights from Stack Overflow and GitHub

T0 review · 4 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read Federated-learning developers' persistent difficulties lie in environment setup, dependency compatibility, API breakage, evaluation, and privacy integration—not in FL algorithm theory—and this paper maps them from 9,611 Stack Overflow and G

desk verdict New empirical taxonomy of FL developer discussions, but sampling makes 'FL-wide' claims overgeneralized: SO is TFF/PySyft traffic, GitHub may be FATE-heavy. read the letter →

arxiv 2607.19621 v1 pith:CZXRKQRL submitted 2026-07-21 cs.SE cs.AI

Sahand Saed , Khairul Alam , Banani Roy This is my paper
classification cs.SEcs.AI
keywords federatedlearningdeveloperchallengesStackOverflowGitHubissuestopicmodelingBERTopicquestionintentanalysissoftwareengineering
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

Federated learning (FL) is hard to build and operate, but the paper asks where the difficulty actually bites. By mining 495 Stack Overflow posts and 9,116 GitHub issues and pull requests from 92 popular FL repositories, it tries to show that the recurring pain points are engineering problems: environment setup and dependency conflicts, API breakages and migration, training instability under non-IID data, evaluation and metric correctness, and integrating privacy mechanisms—not the FL algorithms themselves. It also claims that Stack Overflow functions as a How-to venue (51.11% How-type posts, 3.01x odds of being How versus GitHub) while GitHub serves as a Why-type, root-cause venue (43.86% Why-type artifacts). If true, the result reframes FL adoption barriers as software-maintenance and developer-experience problems, with direct implications for framework design, documentation, and education. The method is repeatable, so the same pipeline could continuously monitor pain points as frameworks evolve.

What carries the argument

The method rests on BERTopic, an embedding-based topic-modeling pipeline that combines sentence-transformer embeddings, UMAP dimensionality reduction, and HDBSCAN clustering, applied independently to the Stack Overflow and GitHub corpora. This is followed by a hybrid rule-based and zero-shot question-intent classifier (validated against manual labels with Cohen's kappa of 0.78) and two resolution-difficulty metrics: the share of posts without an accepted answer or closed artifact, and the median hours to resolution. The work this machinery does is to convert noisy, short developer texts into stable topic clusters, assign each artifact one of four communicative intents (How, Why, What, Other)

What would settle it

Apply the same BERTopic and intent-classification pipeline to Stack Overflow posts that discuss federated training but carry only generic tags (e.g., python, pytorch, tensorflow) and to GitHub repositories with fewer than 300 stars or non-English discussions; if the resulting topic clusters and difficulty rankings shift materially—for example, algorithm-related confusion or collaboration issues rising above setup problems—the paper's concentration-on-engineering claim is falsified. A complementary survey of practitioners working in private or proprietary FL deployments could test the same find

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a two-platform map of FL developer challenges: nine dominant topics on Stack Overflow (TFF implementation, API and compatibility errors, convergence and evaluation mismatch, model state and aggregation, installation and environment compatibility, data partitioning and loading, PySyft setup and API breakages, differential privacy integration, and framework scope and deployment reality) and thirteen on GitHub (deployment and operational failures, data partitioning and pipelines, debugging and configuration, benchmark reproducibility, wiki documentation and collaboration, training instability and evaluation failures, FATE flow operations, feder

Load-bearing premise

The results rest on the premise that the sampled public discussions—Stack Overflow posts tagged with federated-learning plus co-occurring tags passing the 20% TRT and 10% TST thresholds, and GitHub artifacts from 92 English-language repositories with at least 300 stars—are representative of FL developer pain points; if excluded communities, generic-tag posts, smaller repositories, or private channels experience systematically different problems, the topic taxonomy and difficu

Editorial extensions

If this is right

  • Framework maintainers should treat version pinning, dependency compatibility, and migration guides as first-class features, since installation and API breakage topics show among the highest unresolved rates and longest resolution times.
  • Documentation and tutorials should be written as procedural, step-by-step guidance, given that How-type questions dominate Stack Overflow; GitHub's Why-heavy distribution suggests maintainers should also publish root-cause explanations and debugging guides.
  • Evaluation correctness needs standardized utilities: the recurrence of 'runs but does not learn' and misleading metrics points to reference data partitions, seed controls, and validated evaluation pipelines as necessary complements to aggregation algorithms.
  • The repeatable methodology enables low-cost longitudinal monitoring of FL pain points, allowing framework teams to track whether improvements actually reduce unresolved rates and resolution times over time.
  • Educators should pair FL algorithm concepts with hands-on exercises in dependency management, non-IID data preparation, metric aggregation, distributed debugging, and privacy configuration, because the evidence shows these are the real-world blockers.

Reading between the lines

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

  • A natural extension the authors leave implicit is applying the same pipeline to official forums, Slack/Discord channels, mailing lists, and non-English communities; if those channels show different topic mixes, the observed How/Why split may partly reflect platform culture rather than underlying developer needs.
  • The pattern of 0% unresolved but roughly 1,653-hour median resolution for SecureBoost-style issues suggests a support model where fixes depend on maintainers rather than the community; the paper does not test this hypothesis, but its own data make it a concrete prediction.
  • The tag-threshold filter likely excludes beginner questions phrased only with generic tags like 'python' or 'pytorch'; if those posts were included, the share of How-type and setup-related topics could be even higher than reported, or a distinct novice-pain cluster might emerge.
  • If the paper's concentration-on-engineering claim is right, unresolved rate could become a practical quality signal for FL frameworks, analogous to build-failure rates for CI systems—an operational use the authors do not explicitly propose.
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.

Referee Report

4 major / 4 minor

Summary. The paper reports an empirical study of federated learning (FL) developer challenges based on 495 Stack Overflow posts and 9,116 GitHub issues and pull requests from 92 repositories. Using BERTopic topic modeling, the authors derive nine Stack Overflow topics and thirteen GitHub topics; they also classify artifacts into How/Why/What/Other intents and measure topic difficulty via unresolved rates and median resolution times. The main quantitative claims are that Stack Overflow questions are predominantly How-type (51.11%) while GitHub artifacts are predominantly Why-type (43.86%), with OR=3.01 for How on Stack Overflow versus GitHub, and that persistent difficulties concentrate in environment setup, API breakages, non-IID training instability, evaluation correctness, and privacy-preserving integration. The paper concludes with recommendations for framework designers, documentation authors, and educators.

Significance. If the sampling and measurement concerns are adequately addressed, this would be a useful empirical contribution to FL software engineering. The study provides a manually validated taxonomy of developer pain points, a cross-platform comparison of question intent, a reproducible pipeline with a replication package, and explicit discussion of threats to validity. The finding that FL developer discussions center on engineering and maintenance rather than purely algorithmic issues is potentially valuable and actionable. However, the strength of the contribution currently depends on whether the samples are representative of FL developers broadly or only of specific framework communities and large repositories.

major comments (4)
  1. [§3.1.2, Table 1] The TRT>20% and TST>10% criteria admit only `federated-learning`, `tensorflow-federated`, and `pysyft` from Table 1. The resulting 495 SO posts are therefore drawn from support traffic for two frameworks, not a broad sample of FL developer discussions. This is confirmed by Table 2, where six of nine topics are explicitly TFF-specific or PySyft-specific. The abstract's general claim that the study surfaces FL developer pain points dominated by environment setup and API breakages is not supported by the SO data as currently sampled. The authors should either restrict SO-level conclusions to TFF/PySyft contexts or broaden the tag-expansion procedure (e.g., include `flower`, `openfl`, `fedml`, and generic tags with manual screening) and re-run the analysis.
  2. [§3.3, Table 3] The GitHub inclusion criteria do not specify how a repository was identified as FL-related; only popularity, time frame, and language are given. No per-repository artifact counts are reported. Topics 7, 8, 10, and 11 are FATE/KubeFATE-specific, and if the 9,116 issues are concentrated in a small number of projects, the 13-topic taxonomy mainly reflects those projects' issue-tracker activity. Please report the distribution of artifacts across repositories and provide at least a sensitivity analysis excluding the largest repository or examining per-repository topic proportions.
  3. [Abstract and §4.2] The statement that 'How'-type questions dominate is true only for Stack Overflow (51.11%). Across the combined corpus, Why-type artifacts are more numerous (4,069 vs. 2,602), and the paper's own Table 6 shows GitHub is predominantly Why-oriented. The abstract's unqualified claim is misleading and should be revised to a platform-specific statement, e.g., 'How-type questions dominate on Stack Overflow, while Why-type artifacts dominate on GitHub.'
  4. [Table 8, §4.3] Table 8 reports unresolved rates and median resolution times without artifact counts. This makes it impossible to assess reliability; for example, 'Federated Feature Engineering and SecureBoost Issues' has 0% unresolved, but with unknown n this could represent a handful of issues. Since §6 identifies sample-size sensitivity as a validity concern, the table should include n per topic. Additionally, the topic labels in Table 8 should exactly match those in Table 3.
minor comments (4)
  1. [§2.2 and §3.4] The embedding model is described inconsistently: §2.2 says `all-MiniLM-L6-v2`, while §3.4 says `multi-qa-MiniLM-L6-dot-v1`. Please specify the actual model used for the reported experiments.
  2. [Throughout] The manuscript contains numerous typos and formatting issues, e.g., 'T able' in Table 1, 'on on GitHub' in Table 8 title, and inconsistent topic names between Table 3 and Table 8 (e.g., 'FL Deployment and Operational Failures' vs. 'FL System Deployment and Operational Failures'). A careful proofreading pass is needed.
  3. [§3.4] The hyperlink for CountVectorizer is broken ('feature extraction.text.CountVectorizer' with a space). The URL should be corrected.
  4. [§6] The threats-to-validity section is thorough, but the temporal validity discussion could briefly note that GitHub issues closed long after creation may reflect project prioritization rather than technical difficulty, a point already made in §4.3 and worth reinforcing in the threat analysis.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the study is a descriptive empirical analysis; topic taxonomy and difficulty measures are summaries of independently collected data, not fitted parameters or self-citation-derived results.

full rationale

The paper does not derive predictive quantities from fitted constants. RQ1 topics (Tables 2 and 3) are outputs of BERTopic applied to 495 Stack Overflow posts and 9,116 GitHub artifacts, with manual label validation; the abstract's claim that difficulties concentrate in setup/API/evaluation is a frequency and difficulty summary of those clusters, not an equation whose output equals its input. RQ2's How/Why distribution comes from a rule-based plus zero-shot classifier with manual validation (Cohen's kappa = 0.78) and standard chi-square/odds-ratio calculations; no fitted parameter is renamed as a prediction. RQ3 difficulty metrics are raw metadata (accepted-answer status, closure status, median resolution times). The self-citations [19,23,24] are related-work examples for topic modeling on GitHub and are not load-bearing: they do not supply a uniqueness theorem or constrain the taxonomy. The strongest potential concern, identified by the skeptic and partly acknowledged in Section 6, is sampling and construct validity: the TRT/TST tag filter in Section 3.1.2 admits only federated-learning, tensorflow-federated, and pysyft, so the Stack Overflow corpus is TFF/PySyft-heavy, and the GitHub topics may be repo-dominated. This is a limitation on generalizability, not a circular reduction: the corpus composition is not definitionally identical to the conclusions, and the paper explicitly frames its recommendations as evidence-informed rather than validated best practices in Section 5.2. Therefore no circular step meeting the required quote-and-reduction standard is present.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The ledger reflects the sampling and modeling choices the taxonomy depends on. No invented physical or conceptual entities are introduced. The central descriptive claims do not rest on mathematical constants or new theoretical objects.

free parameters (6)
  • TRT tag relevance threshold = 20%
    Hand-chosen by three raters to include FL-related tags while filtering noise; directly controls the 495-post SO corpus (§3.1.2).
  • TST tag significance threshold = 10%
    Hand-chosen cutoff for co-occurrence frequency with federated-learning; also controls SO corpus (§3.1.2).
  • BERTopic UMAP n_neighbors = 30 (SO), 20 (GitHub)
    Selected by comparing topic coherence scores on the same datasets (§3.4); different values change topic structure.
  • BERTopic UMAP n_components = 3 (SO), 4 (GitHub)
    Tuned jointly with n_neighbors and min_cluster_size to maximize coherence (§3.4).
  • HDBSCAN min_cluster_size = 210 (SO), 130 (GitHub)
    Tuned to balance topic granularity and coherence; directly affects how many topics emerge (§3.4).
  • How/Why/What classifier thresholds = rule thresholds in unverified script
    Rule-based confidence thresholds and zero-shot fallback criteria are pipeline hyperparameters; exact values are only in the replication-package script.
assumptions (5)
  • domain assumption The sampled SO tags and 300-star GitHub repos represent FL developer pain points.
    Central sampling premise; if excluded discussions differ, the taxonomy is not general (§3.1.2, §3.3, §6).
  • domain assumption BERTopic clusters correspond to meaningful developer challenge topics.
    No automatic ground truth; labels are assigned by authors after manual inspection (§3.4, §6).
  • domain assumption No accepted answer or open/unclosed state indicates difficulty.
    Authors call these indirect indicators; closure can mean duplicate, abandoned, or invalid rather than resolved (§4.3, §6).
  • domain assumption Median resolution time among closed artifacts is interpretable as difficulty.
    Right-censoring is acknowledged; long times may reflect maintainer scheduling, release cycles, or backlog rather than complexity (§6).
  • domain assumption How/Why/What taxonomy transfers from Stack Overflow Q&A to GitHub issues and PRs.
    GitHub artifacts include bug reports, feature requests, and coordination; the paper uses Other to absorb this but acknowledges simplification (§4.2, §6).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Understanding Developer Pain Points in Federated Learning: Insights from Stack Overflow and GitHub." pith.science (2026). https://pith.science/paper/CZXRKQRL

@misc{pith2026260719621,
  author       = {Pith},
  title        = {Pith review of: Understanding Developer Pain Points in Federated Learning: Insights from Stack Overflow and GitHub},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CZXRKQRL}},
  note         = {Machine review of arXiv:2607.19621}
}
read the original abstract

Federated Learning (FL) enables collaborative model training without centralizing raw data, but building and operating FL systems remains difficult due to distributed execution, rapidly evolving frameworks, and privacy and governance requirements. In this paper, we present an empirical study of FL developer challenges by independently analyzing 495 Stack Overflow posts and 9,116 GitHub issues and pull requests from 92 FL-related projects. Using BERTopic-based topic modeling and difficulty indicators such as unresolved rates and median resolution time, we characterize recurring problem areas and compare how they manifest across the two support platforms, Stack Overflow and GitHub. Our analysis surfaces nine dominant Stack Overflow topics and thirteen GitHub topics, with persistent difficulties concentrated in environment setup and dependency compatibility, API breakages and migration, training instability under non-IID data, evaluation and metric correctness, and the integration of privacy-preserving mechanisms. We also categorize posts by question intent to understand the kinds of help developers seek; this intent analysis shows that "How"-type questions dominate, reflecting strong demand for procedural guidance. Several topics, such as "TFF Installation and Environment Compatibility" and "Federated Feature Engineering and SecureBoost Issues," exhibit high unresolved rates and long resolution times, suggesting shortcomings in tooling, documentation, and debugging support. Based on these findings, we provide actionable implications for FL framework designers, documentation authors, and educators. Although our results are constrained to public discussions and a subset of widely discussed frameworks, the study offers a scalable method for continuously monitoring developer pain points and improving the usability, reliability, and deployability of FL systems.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

96 extracted references · 1 canonical work pages

  1. [1]

    Medical image analysis42, 60–88 (2017)

    Litjens, G., Kooi, T., Bejnordi, B.E., Setio, A.A.A., Ciompi, F., Ghafoorian, M., Van Der Laak, J.A., Van Ginneken, B., S´ anchez, C.I.: A survey on deep learning in medical image analysis. Medical image analysis42, 60–88 (2017)

    2017

  2. [2]

    Off J Eur Union119, 1–88 (2016)

    Parliament, E., Council: Regulation (eu) 2016/679 (general data protection regulation). Off J Eur Union119, 1–88 (2016)

    2016

  3. [3]

    Information & communications technology law28(1), 65–98 (2019)

    Hoofnagle, C.J., Van Der Sloot, B., Borgesius, F.Z.: The european union general data protection regulation: what it is and what it means. Information & communications technology law28(1), 65–98 (2019)

    2019

  4. [4]

    In: Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS) (2017)

    McMahan, H.B., Moore, E., Ramage, D., Hampson, S., Arcas, B.: Communication-efficient learning of deep networks from decentralized data. In: Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS) (2017)

    2017

  5. [5]

    Proceedings of machine learning and systems1, 374–388 (2019)

    Bonawitz, K., Eichner, H., Grieskamp, W., Huba, D., Ingerman, A., Ivanov, V., Kiddon, C., Koneˇ cn` y, J., Mazzocchi, S., McMahan, B.,et al.: Towards federated learning at scale: System design. Proceedings of machine learning and systems1, 374–388 (2019)

    2019

  6. [6]

    arXiv preprint arXiv:1806.00582 (2018)

    Zhao, Y., Li, M., Lai, L., Suda, N., Civin, D., Chandra, V.: Federated learning with non-iid data. arXiv preprint arXiv:1806.00582 (2018)

    arXiv 2018

  7. [7]

    IEEE Signal Processing Magazine37(3), 50–60 (2020)

    Li, T., Sahu, A.K., Talwalkar, A., Smith, V.: Federated learning: Challenges, methods, and future directions. IEEE Signal Processing Magazine37(3), 50–60 (2020)

    2020

  8. [8]

    Foundations and Trends in Machine Learning14(1–2), 1–210 (2021) https://doi.org/10.1561/2200000083

    Kairouz, P., McMahan, H.B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A.N.,et al.: Advances and open problems in federated learning. Foundations and Trends in Machine Learning14(1–2), 1–210 (2021) https://doi.org/10.1561/2200000083

    doi:10.1561/2200000083 2021

Show all 96 references
  1. [9]

    arXiv preprint arXiv:1901.11173 (2019)

    Lalitha, A., Kilinc, O.C., Javidi, T., Koushanfar, F.: Peer-to-peer federated learning on graphs. arXiv preprint arXiv:1901.11173 (2019)

    1901 arXiv

  2. [10]

    Journal of Parallel and Distributed Computing148, 109–124 (2021)

    Heged˝ us, I., Danner, G., Jelasity, M.: Decentralized learning works: An empirical comparison of gossip learning and federated learning. Journal of Parallel and Distributed Computing148, 109–124 (2021)

    2021

  3. [11]

    Scientific Reports10(1), 12598 (2020) https://doi.org/10.1038/s41598-020-69250-1

    Sheller, M.J., Edwards, B., Reina, G.A., Martin, J., Pati, S., Kotrotsou, A., Milchenko, M., Xu, W., Marcus, D., Colen, R.R., Bakas, S.: Federated learning in medicine: facilitating multi- institutional collaborations without sharing patient data. Scientific Reports10(1), 1259...

    2020 doi

  4. [12]

    Sensors 21(1), 167 (2020)

    Kholod, I., Yanaki, E., Fomichev, D., Shalugin, E., Novikova, E., Filippov, E., Nordlund, M.: Open-source federated learning frameworks for iot: A comparative review and analysis. Sensors 21(1), 167 (2020)

    2020

  5. [13]

    In: Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security (CCS), pp

    Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, H.B., Patel, S., Ramage, D., Segal, A., Seth, K.: Practical secure aggregation for privacy-preserving machine learning. In: Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security (CCS...

    2017

  6. [14]

    In: Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security (CCS), pp

    Abadi, M., Chu, A., Goodfellow, I., McMahan, H.B., Mironov, I., Talwar, K., Zhang, L.: Deep learning with differential privacy. In: Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security (CCS), pp. 308–318 (2016). https://doi.org/10.1145/ 2976749.2978318

    2016

  7. [17]

    arXiv preprint (2021) arXiv:2105.06413

    Reina, G.A., et al.: Openfl: An open-source framework for federated learning. arXiv preprint (2021) arXiv:2105.06413

    2021 arXiv

  8. [18]

    In: Proceedings of the 27th ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA), pp

    Zhang, Y., Chen, Y., Cheung, S.-C., Xiong, Y., Qin, L.: An empirical study on tensorflow program bugs. In: Proceedings of the 27th ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA), pp. 129–140 (2018). https://doi.org/10.1145/3213846.3213866

    2018

  9. [19]

    arXiv preprint arXiv:2601.09612 (2026)

    Alam, K., Roy, B.: Analyzing github issues and pull requests in nf-core pipelines: Insights into nf-core pipeline repositories. arXiv preprint arXiv:2601.09612 (2026)

    2026

  10. [20]

    In: Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE), pp

    Humbatova, N., Jahangirova, G., Bavota, G., Riccio, V., Stocco, A., Tonella, P.: Taxonomy of real faults in deep learning systems. In: Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering (ICSE), pp. 1110–1121 (2020). https://doi.org/10.1145/ 33778...

    2020

  11. [21]

    In: 2019 ACM/IEEE Inter- national Symposium on Empirical Software Engineering and Measurement (ESEM), pp

    Alshangiti, M., Sapkota, H., Murukannaiah, P.K., Liu, X., Yu, Q.: Why is developing machine learning applications challenging? a study on stack overflow posts. In: 2019 ACM/IEEE Inter- national Symposium on Empirical Software Engineering and Measurement (ESEM), pp. 1–11 (2019)...

    2019

  12. [22]

    In: 2021 IEEE 21st International Working Conference on Source Code Analysis and Manipulation (SCAM), pp

    Hamidi, A., Antoniol, G., Khomh, F., Di Penta, M., Hamidi, M.: Towards understanding devel- opers’ machine-learning challenges: A multi-language study on stack overflow. In: 2021 IEEE 21st International Working Conference on Source Code Analysis and Manipulation (SCAM), pp. 58...

    2021

  13. [23]

    Empirical Software Engineering30(5), 151 (2025)

    Alam, K., Roy, B., Roy, C.K., Mittal, K.: An empirical investigation on the challenges in scientific workflow systems development. Empirical Software Engineering30(5), 151 (2025)

    2025

  14. [24]

    arXiv preprint arXiv:2607.10839 (2026)

    Alam, K., Roy, K., Rahman, M.S., Roy, B.: Maintenance and support in community-driven scientific pipeline ecosystems: A cross-platform empirical study of nf-core. arXiv preprint arXiv:2607.10839 (2026)

    2026 arXiv

  15. [25]

    Journal of healthcare informatics research5, 1–19 (2021)

    Xu, J., Glicksberg, B.S., Su, C., Walker, P., Bian, J., Wang, F.: Federated learning for healthcare informatics. Journal of healthcare informatics research5, 1–19 (2021)

    2021

  16. [26]

    IEEE communications surveys & tutorials23(3), 1622–1658 (2021)

    Nguyen, D.C., Ding, M., Pathirana, P.N., Seneviratne, A., Li, J., Poor, H.V.: Federated learning for internet of things: A comprehensive survey. IEEE communications surveys & tutorials23(3), 1622–1658 (2021)

    2021

  17. [27]

    ACM Computing Surveys (Csur)55(3), 1–37 (2022)

    Nguyen, D.C., Pham, Q.-V., Pathirana, P.N., Ding, M., Seneviratne, A., Lin, Z., Dobre, O., Hwang, W.-J.: Federated learning for smart healthcare: A survey. ACM Computing Surveys (Csur)55(3), 1–37 (2022)

    2022

  18. [28]

    Journal of Network and Computer Applications, 103714 (2023) 39

    Guendouzi, B.S., Ouchani, S., Assaad, H.E., Zaher, M.E.: A systematic review of federated learning: Challenges, aggregation methods, and development tools. Journal of Network and Computer Applications, 103714 (2023) 39

    2023

  19. [29]

    IEEE communications surveys & tutorials22(3), 2031–2063 (2020)

    Lim, W.Y.B., Luong, N.C., Hoang, D.T., Jiao, Y., Liang, Y.-C., Yang, Q., Niyato, D., Miao, C.: Federated learning in mobile edge networks: A comprehensive survey. IEEE communications surveys & tutorials22(3), 2031–2063 (2020)

  20. [30]

    IEEE Access (2023)

    Vucinich, S., Zhu, Q.: The current state and challenges of fairness in federated learning. IEEE Access (2023)

    2023

  21. [31]

    arXiv preprint arXiv:2403.06067 (2024)

    Shirvani, G., Ghasemshirazi, S., Beigzadeh, B.: Federated learning: Attacks, defenses, opportu- nities, and challenges. arXiv preprint arXiv:2403.06067 (2024)

    2024 arXiv

  22. [32]

    Multimedia tools and applications78, 15169–15211 (2019)

    Jelodar, H., Wang, Y., Yuan, C., Feng, X., Jiang, X., Li, Y., Zhao, L.: Latent dirichlet allocation (lda) and topic modeling: models, applications, a survey. Multimedia tools and applications78, 15169–15211 (2019)

    2019

  23. [33]

    In: Advances in Information Retrieval: 31th European Conference on IR Research, ECIR 2009, Toulouse, France, April 6-9, 2009

    Yi, X., Allan, J.: A comparative study of utilizing topic models for information retrieval. In: Advances in Information Retrieval: 31th European Conference on IR Research, ECIR 2009, Toulouse, France, April 6-9, 2009. Proceedings 31, pp. 29–41 (2009). Springer

    2009

  24. [34]

    University of Massachusetts Amherst, ??? (2007)

    Wei, X.: Topic Models in Information Retrieval. University of Massachusetts Amherst, ??? (2007)

    2007

  25. [35]

    Scientometrics100, 767–786 (2014)

    Yau, C.-K., Porter, A., Newman, N., Suominen, A.: Clustering scientific documents with topic modeling. Scientometrics100, 767–786 (2014)

    2014

  26. [36]

    arXiv preprint arXiv:1309.6874 (2013)

    Xie, P., Xing, E.P.: Integrating document clustering and topic modeling. arXiv preprint arXiv:1309.6874 (2013)

    2013 arXiv

  27. [37]

    Journal of Intelligence and Information systems21(3), 101–116 (2015)

    Choi, S., Hyun, Y., Kim, N.: Improving performance of recommendation systems using topic modeling. Journal of Intelligence and Information systems21(3), 101–116 (2015)

    2015

  28. [38]

    In: Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013), pp

    Luostarinen, T., Kohonen, O.: Using topic models in content-based news recommender systems. In: Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013), pp. 239–251 (2013)

    2013

  29. [39]

    In: Web Information Systems and Technologies: 10th International Conference, WEBIST 2014, Barcelona, Spain, April 3-5, 2014, Revised Selected Papers 10, pp

    Bergamaschi, S., Po, L.: Comparing lda and lsa topic models for content-based movie recommen- dation systems. In: Web Information Systems and Technologies: 10th International Conference, WEBIST 2014, Barcelona, Spain, April 3-5, 2014, Revised Selected Papers 10, pp. 247–263 (2...

    2014

  30. [40]

    Soft Computing27(7), 3965–3982 (2023)

    Belwal, R.C., Rai, S., Gupta, A.: Extractive text summarization using clustering-based topic modeling. Soft Computing27(7), 3965–3982 (2023)

    2023

  31. [41]

    In: Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pp

    Eidelman, V., Boyd-Graber, J., Resnik, P.: Topic models for dynamic translation model adap- tation. In: Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pp. 115–119 (2012)

    2012

  32. [42]

    In: Proceedings of the 32nd ACM/IEEE International Conference on Software Engineering-Volume 1, pp

    Asuncion, H.U., Asuncion, A.U., Taylor, R.N.: Software traceability with topic modeling. In: Proceedings of the 32nd ACM/IEEE International Conference on Software Engineering-Volume 1, pp. 95–104 (2010)

    2010

  33. [43]

    In: Advances in Knowledge Discovery and Data Mining: 15th Pacific-Asia Conference, PAKDD 2011, Shenzhen, China, May 24-27, 2011, Proceedings, Part I 15, pp

    Zhai, Z., Liu, B., Xu, H., Jia, P.: Constrained lda for grouping product features in opinion mining. In: Advances in Knowledge Discovery and Data Mining: 15th Pacific-Asia Conference, PAKDD 2011, Shenzhen, China, May 24-27, 2011, Proceedings, Part I 15, pp. 448–459 (2011). Springer

    2011

  34. [44]

    In: 2010 IEEE International Conference on Software Maintenance, pp

    Gethers, M., Poshyvanyk, D.: Using relational topic models to capture coupling among classes in object-oriented software systems. In: 2010 IEEE International Conference on Software Maintenance, pp. 1–10 (2010). IEEE

    2010

  35. [45]

    In: 2024 12th International Symposium on Digital Forensics and Security (ISDFS), pp

    Gokcimen, T., Das, B.: Topic modelling using bertopic for robust spam detection. In: 2024 12th International Symposium on Digital Forensics and Security (ISDFS), pp. 1–5 (2024). IEEE 40

    2024

  36. [46]

    Journal of machine Learning research3(Jan), 993–1022 (2003)

    Blei, D.M., Ng, A.Y., Jordan, M.I.: Latent dirichlet allocation. Journal of machine Learning research3(Jan), 993–1022 (2003)

    2003

  37. [47]

    Advances in neural information processing systems 18, 147 (2006)

    Blei, D., Lafferty, J.: Correlated topic models. Advances in neural information processing systems 18, 147 (2006)

    2006

  38. [48]

    In: Proceedings of the 23rd International Conference on Machine Learning, pp

    Blei, D.M., Lafferty, J.D.: Dynamic topic models. In: Proceedings of the 23rd International Conference on Machine Learning, pp. 113–120 (2006)

    2006

  39. [49]

    Advances in neural information processing systems13(2000)

    Lee, D., Seung, H.S.: Algorithms for non-negative matrix factorization. Advances in neural information processing systems13(2000)

    2000

  40. [50]

    In: Proceedings of the 22nd International Conference on World Wide Web, pp

    Yan, X., Guo, J., Lan, Y., Cheng, X.: A biterm topic model for short texts. In: Proceedings of the 22nd International Conference on World Wide Web, pp. 1445–1456 (2013)

    2013

  41. [51]

    arXiv preprint arXiv:2203.05794 (2022)

    Grootendorst, M.: Bertopic: Neural topic modeling with a class-based tf-idf procedure. arXiv preprint arXiv:2203.05794 (2022)

    2022 arXiv

  42. [52]

    arXiv preprint arXiv:1908.10084 (2019)

    Reimers, N., Gurevych, I.: Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084 (2019)

    1908 arXiv

  43. [53]

    Hugging Face model card (2026)

    Sentence-Transformers: all-MiniLM-L6-v2. Hugging Face model card (2026). https:// huggingface.co/sentence-transformers/all-MiniLM-L6-v2 Accessed 2026-01-03

    2026

  44. [54]

    arXiv preprint arXiv:1802.03426 (2018)

    McInnes, L., Healy, J., Melville, J.: Umap: Uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426 (2018)

    2018 arXiv

  45. [55]

    the Journal of machine Learning research12, 2825–2830 (2011)

    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V.,et al.: Scikit-learn: Machine learning in python. the Journal of machine Learning research12, 2825–2830 (2011)

    2011

  46. [56]

    https://scikit-learn.org/stable/modules/generated/sklearn.feature extraction.text

    scikit-learn developers: sklearn.feature extraction.text.CountVectorizer documentation (2026). https://scikit-learn.org/stable/modules/generated/sklearn.feature extraction.text. CountVectorizer.html Accessed 2026-01-03

    2026

  47. [57]

    In: International Conference on Human-Computer Interaction, pp

    Chen, L.: What do user experience professionals discuss online? topic modeling of a user experience q&a community. In: International Conference on Human-Computer Interaction, pp. 365–380 (2023). Springer

    2023

  48. [58]

    Frontiers of computer science in China4, 280–301 (2010)

    Daud, A., Li, J., Zhou, L., Muhammad, F.: Knowledge discovery through directed probabilistic topic models: a survey. Frontiers of computer science in China4, 280–301 (2010)

    2010

  49. [59]

    IEEE transactions on knowledge and data engineering29(6), 1186–1198 (2017)

    Nie, L., Wei, X., Zhang, D., Wang, X., Gao, Z., Yang, Y.: Data-driven answer selection in community qa systems. IEEE transactions on knowledge and data engineering29(6), 1186–1198 (2017)

    2017

  50. [60]

    In: Topic Detection and Tracking: Event-based Information Organization, pp

    Fiscus, J.G., Doddington, G.R.: Topic detection and tracking evaluation overview. In: Topic Detection and Tracking: Event-based Information Organization, pp. 17–31. Springer, ??? (2002)

    2002

  51. [61]

    Empirical Software Engineering21, 1192–1223 (2016)

    Rosen, C., Shihab, E.: What are mobile developers asking about? a large scale study using stack overflow. Empirical Software Engineering21, 1192–1223 (2016)

    2016

  52. [62]

    In: 2019 ACM/IEEE Inter- national Symposium on Empirical Software Engineering and Measurement (ESEM), pp

    Alshangiti, M., Sapkota, H., Murukannaiah, P.K., Liu, X., Yu, Q.: Why is developing machine learning applications challenging? a study on stack overflow posts. In: 2019 ACM/IEEE Inter- national Symposium on Empirical Software Engineering and Measurement (ESEM), pp. 1–11 (2019). IEEE

    2019

  53. [63]

    In: Proceedings of the 12th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement, pp

    Ahmed, S., Bagherzadeh, M.: What do concurrency developers ask about? a large-scale study using stack overflow. In: Proceedings of the 12th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement, pp. 1–10 (2018) 41

    2018

  54. [64]

    Journal of Computer Science and Technology31, 910–924 (2016)

    Yang, X.-L., Lo, D., Xia, X., Wan, Z.-Y., Sun, J.-L.: What security questions do developers ask? a large-scale study of stack overflow posts. Journal of Computer Science and Technology31, 910–924 (2016)

    2016

  55. [65]

    In: 2023 49th Euromicro Conference on Software Engineering and Advanced Applications (SEAA), pp

    Mi, Q., Bao, Q., Cui, L.: Identifying topics and trends in devops: A study of stack overflow posts. In: 2023 49th Euromicro Conference on Software Engineering and Advanced Applications (SEAA), pp. 402–409 (2023). IEEE

    2023

  56. [66]

    In: Proceedings of the 17th International Conference on Mining Software Repositories, pp

    Abdellatif, A., Costa, D., Badran, K., Abdalkareem, R., Shihab, E.: Challenges in chatbot devel- opment: A study of stack overflow posts. In: Proceedings of the 17th International Conference on Mining Software Repositories, pp. 174–185 (2020)

    2020

  57. [67]

    Information and Software Technology84, 19–32 (2017)

    Zou, J., Xu, L., Yang, M., Zhang, X., Yang, D.: Towards comprehending the non-functional requirements through developers’ eyes: An exploration of stack overflow using topic analysis. Information and Software Technology84, 19–32 (2017)

    2017

  58. [68]

    In: 2021 IEEE International Conference on Software Maintenance and Evolution (ICSME), pp

    Li, H., Khomh, F., Openja, M.,et al.: Understanding quantum software engineering challenges an empirical study on stack exchange forums and github issues. In: 2021 IEEE International Conference on Software Maintenance and Evolution (ICSME), pp. 343–354 (2021). IEEE

    2021

  59. [69]

    In: 2021 IEEE/ACM 18th International Conference on Mining Software Repositories (MSR), pp

    Scoccia, G.L., Migliarini, P., Autili, M.: Challenges in developing desktop web apps: a study of stack overflow and github. In: 2021 IEEE/ACM 18th International Conference on Mining Software Repositories (MSR), pp. 271–282 (2021). IEEE

    2021

  60. [70]

    In: Proceedings of the 13th Innovations in Software Engineering Conference (formerly Known as India Software Engineering Conference), pp

    Dhasade, A.B., Venigalla, A.S.M., Chimalakonda, S.: Towards prioritizing github issues. In: Proceedings of the 13th Innovations in Software Engineering Conference (formerly Known as India Software Engineering Conference), pp. 1–5 (2020)

    2020

  61. [71]

    Jokhio, M.: Mining github issues for bugs, feature requests and questions (2021)

    2021

  62. [72]

    In: The Art and Science of Analyzing Software Data, pp

    Campbell, J.C., Hindle, A., Stroulia, E.: Latent dirichlet allocation: extracting topics from soft- ware engineering data. In: The Art and Science of Analyzing Software Data, pp. 139–159. Elsevier, ??? (2015)

    2015

  63. [73]

    94–97 (2019)

    Wang, X., Lee, M., Pinchbeck, A., Fard, F.: Where does lda sit for github? In: 2019 34th IEEE/ACM International Conference on Automated Software Engineering Workshop (ASEW), pp. 94–97 (2019). IEEE

    2019

  64. [74]

    Empirical Software Engineering25, 2694–2747 (2020) https://doi.org/10

    Han, J., Shihab, E., Wan, Z., Deng, S., Xia, X.: What do programmers discuss about deep learning frameworks. Empirical Software Engineering25, 2694–2747 (2020) https://doi.org/10. 1007/s10664-020-09819-6

    2020

  65. [75]

    Empirical software engineering19, 619–654 (2014)

    Barua, A., Thomas, S.W., Hassan, A.E.: What are developers talking about? an analysis of topics and trends in stack overflow. Empirical software engineering19, 619–654 (2014)

    2014

  66. [76]

    In: Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pp

    Bagherzadeh, M., Khatchadourian, R.: Going big: a large-scale study on what big data develop- ers ask. In: Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pp. 432–442 (2019)

    2019

  67. [77]

    IEEE Transactions on Software Engineering47(7), 1331–1349 (2019)

    Wan, Z., Xia, X., Hassan, A.E.: What do programmers discuss about blockchain? a case study on the use of balanced lda and the reference architecture of a domain to capture online discussions about blockchain platforms across stack exchange communities. IEEE Transactions on Sof...

    2019

  68. [78]

    Developer Challenges on Federated Learning

    Saed, S., co-authors: FL Developers Challenges: Replication Package for “Developer Challenges on Federated Learning”. https://github.com/sahandsaed/FL Developers Challenges. GitHub repository (2026)

    2026

  69. [79]

    In: Proceedings of the COLING/ACL 2006 Interactive Presentation Sessions, pp

    Bird, S.: Nltk: the natural language toolkit. In: Proceedings of the COLING/ACL 2006 Interactive Presentation Sessions, pp. 69–72 (2006) 42

    2006

  70. [80]

    Packt Publishing Ltd, ??? (2016)

    Hardeniya, N., Perkins, J., Chopra, D., Joshi, N., Mathur, I.: Natural Language Processing: Python and NLTK. Packt Publishing Ltd, ??? (2016)

    2016

  71. [81]

    No Starch Press, ??? (2020)

    Vasiliev, Y.: Natural Language Processing with Python and spaCy: A Practical Introduction. No Starch Press, ??? (2020)

    2020

  72. [82]

    jumping through hoops

    Mezini, S.N.S.K.M., Bodden, E.: “jumping through hoops”: Why do java developers struggle with cryptography apis? GI-Edition, 57

  73. [83]

    https://huggingface.co/models

    Community, H.F.: Hugging Face Models. https://huggingface.co/models. Accessed: 2025-10-07 (2025)

    2025

  74. [84]

    McInnes, L., Healy, J., Astels, S.,et al.: hdbscan: Hierarchical density based clustering. J. Open Source Softw.2(11), 205 (2017)

    2017

  75. [85]

    Philo- sophical transactions of the royal society A: Mathematical, Physical and Engineering Sciences 374(2065), 20150202 (2016)

    Jolliffe, I.T., Cadima, J.: Principal component analysis: a review and recent developments. Philo- sophical transactions of the royal society A: Mathematical, Physical and Engineering Sciences 374(2065), 20150202 (2016)

  76. [86]

    In: Proceedings of the 11th Working Conference on Mining Software Repositories, pp

    Bajaj, K., Pattabiraman, K., Mesbah, A.: Mining questions asked by web developers. In: Proceedings of the 11th Working Conference on Mining Software Repositories, pp. 112–121 (2014)

    2014

  77. [87]

    Treude, C., Barzilay, O., Storey, M.-A.: How do programmers ask and answer questions on the web? In: Proceedings of the 33rd International Conference on Software Engineering (ICSE ’11), pp. 804–807. Association for Computing Machinery (ACM), Waikiki, Honolulu, HI, USA (2011). ...

    2011

  78. [88]

    arXiv preprint arXiv:2007.14390 (2020)

    Beutel, D.J., Topal, T., Mathur, A., Qiu, X., Fernandez-Marques, J., Gao, Y., Sani, L., Li, K.H., Parcollet, T., Gusm˜ ao, P.P.B.d., Lane, N.D.: Flower: A friendly federated learning research framework. arXiv preprint arXiv:2007.14390 (2020)

    2007 arXiv

  79. [89]

    arXiv preprint arXiv:2007.13518 (2020)

    He, C., Annavaram, M., Avestimehr, S., et al.: Fedml: A research library and benchmark for federated machine learning. arXiv preprint arXiv:2007.13518 (2020)

    2007 arXiv

  80. [90]

    Journal of Machine Learning Research22(226), 1–6 (2021)

    Liu, Y., Fan, T., Chen, T., Xu, Q., Yang, Q.: Fate: An industrial grade platform for collaborative learning with data protection. Journal of Machine Learning Research22(226), 1–6 (2021)

    2021

  81. [91]

    arXiv preprint arXiv:2105.06413 (2021) https://doi.org/10.48550/arXiv

    Reina, G.A., Gruzdev, A., Foley, P., Perepelkina, O.,et al.: Openfl: An open-source framework for federated learning. arXiv preprint arXiv:2105.06413 (2021) https://doi.org/10.48550/arXiv. 2105.06413

    2021 doi arXiv

  82. [92]

    npj Digital Medicine3, 119 (2020) https://doi

    Rieke, N., Hancox, J., Li, W., Milletari, F., Roth, H.R., Albarqouni, S., Bakas, S.,et al.: The future of digital health with federated learning. npj Digital Medicine3, 119 (2020) https://doi. org/10.1038/s41746-020-00323-1

    2020 doi

  83. [93]

    In: 2012 International Conference on Software and System Process (ICSSP), pp

    Fagerholm, F., M¨ unch, J.: Developer experience: Concept and definition. In: 2012 International Conference on Software and System Process (ICSSP), pp. 73–77 (2012). IEEE

    2012

  84. [94]

    In: Proceedings of the ACM SIGSOFT 20th International Symposium on the Foundations of Software Engineering, pp

    Robbes, R., Lungu, M., R¨ othlisberger, D.: How do developers react to api deprecation? the case of a smalltalk ecosystem. In: Proceedings of the ACM SIGSOFT 20th International Symposium on the Foundations of Software Engineering, pp. 1–11 (2012)

    2012

  85. [95]

    In: Software Ecosystems: Tooling and Analytics, pp

    Mens, T., Roover, C.D.: An introduction to software ecosystems. In: Software Ecosystems: Tooling and Analytics, pp. 1–29. Springer, ??? (2023)

    2023

  86. [96]

    In: Pro- ceedings of the Second ACM-IEEE International Symposium on Empirical Software Engineering and Measurement, pp

    Cataldo, M., Herbsleb, J.D., Carley, K.M.: Socio-technical congruence: a framework for assessing the impact of technical and work dependencies on software development productivity. In: Pro- ceedings of the Second ACM-IEEE International Symposium on Empirical Software Engineeri...

    2008

  87. [97]

    Organizational Research Methods10(2), 393 (2007)

    Bean, C.J.: Qualitative research design: An interactive approach. Organizational Research Methods10(2), 393 (2007)

    2007

  88. [98]

    Springer, ??? (2012) 44

    Wohlin, C., Runeson, P., H¨ ost, M., Ohlsson, M.C., Regnell, B., Wessl´ en, A.: Experimentation in Software Engineering. Springer, ??? (2012) 44

    2012

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.