Pith. sign in

REVIEW 3 major objections 3 minor 41 references

A laser SLAM method represents the environment as Gaussian-process object contours, updated recursively and inferred jointly with the robot pose in a fully Bayesian framework.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

A laser SLAM framework that models each object as a Gaussian-process contour, updated recursively and inferred jointly with the robot pose in a Bayesian framework.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection The manuscript body is a different paper entirely; the GPL-SLAM abstract has no supporting content, so there is nothing to review. the 3 major comments →

arxiv 2508.16459 v1 pith:Z4CNF34G submitted 2025-08-22 cs.RO

GPL-SLAM: A Laser SLAM Framework with Gaussian Process Based Extended Landmarks

Ali Emre Balc{\i} (1) , Erhan Ege Keyvan (2) , Emre \"Ozkan (2) ((1) TU Delft , (2) Middle East Technical University) This is my paper
classification cs.RO
keywords Gaussian processSLAMobject-based mappinglaser scanBayesian inferencecontour representationprobabilistic data associationsemantic mapping
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 aims to establish that laser-based simultaneous localization and mapping (SLAM) can be built on object-level contour representations modeled by Gaussian processes, instead of conventional grid maps or point-cloud registration. The authors claim that such a representation supports online recursive updates, efficient memory use, and fully Bayesian joint inference over the robot pose and the object map. If correct, a robot could localize and map accurately while storing only compact object contours, and the map would naturally include semantic information such as object counts, areas, and shape-confidence bounds. The authors validate the approach on synthetic and real-world structured environments and report accurate localization and mapping performance.

Core claim

The central claim is that a complete SLAM system can be constructed around Gaussian-process (GP) representations of object contours rather than grid maps or dense point clouds. The environment is modeled on a per-object basis: each object's 2D outline is a GP contour that is updated online through a recursive scheme. The entire SLAM problem is cast in a fully Bayesian framework, enabling joint inference over the robot pose and the object-level map, including probabilistic measurement-to-object associations. Because the map is object-based, the system yields semantic quantities (object count, areas) and GP-derived confidence bounds on object shapes, which are useful for downstream tasks like

What carries the argument

The central object is the Gaussian-process contour: a nonparametric distribution over a closed 2D curve representing a detected object's outline. The method maintains a GP contour for each object and updates it recursively as new laser scans arrive. The load-bearing mechanism is the joint Bayesian inference over the robot pose and the set of object contours, which also produces probabilistic measurement-to-object associations and shape-confidence bounds.

Load-bearing premise

The whole system depends on the environment being decomposable into distinct objects whose 2D contours the laser can observe; where segmentation fails or objects are heavily occluded, the GP contours, recursive updates, and joint inference all lose their footing.

What would settle it

Run the method in a tightly cluttered scene where objects touch or partially hide one another, then compare the reported object count and contour confidence bounds against ground truth; if the recursive updates cause contours to drift or the probabilistic associations misassign scans, the central claim of accurate object-level mapping is called into question.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Robot memory usage scales with the number of objects and their contour complexity, not with grid resolution or point-cloud density.
  • The map output naturally includes semantic summaries—object counts and areas—as byproducts of the contour representation.
  • Shape-confidence bounds from the GP can be used directly to plan safe paths and decide where to explore next.
  • Probabilistic association between laser measurements and objects could handle partial occlusion and object ambiguity more gracefully than hard data-association schemes.
  • If the method holds, object-level maps become practical for long-term autonomy in structured indoor and outdoor environments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The supplied full text does not match the paper's title and abstract; this extraction is based on the abstract and associated metadata only.
  • The same object-contour representation could plausibly be extended to dynamic objects by letting the GP contours evolve over time, though the paper does not claim this.
  • A natural stress test is whether the recursive GP updates remain stable when an object is observed from widely varying viewpoints with heavy occlusion.
  • The approach could be combined with semantic classifiers to tag object contours with categories, yielding a richer map without changing the inference core.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The submission presents an abstract for a paper titled 'GPL-SLAM: A Laser SLAM Framework with Gaussian Process Based Extended Landmarks,' which claims a novel SLAM method using GP-based contour landmarks, recursive online contour updates, fully Bayesian joint inference over robot pose and object map, and validation on synthetic and real-world experiments. The full text that follows, however, is an unrelated software-engineering paper, 'Using LLMs and Essence to Support Software Practice Adoption' (arXiv:2508.16445v1, cs.SE), which develops and evaluates an LLM chatbot with retrieval-augmented generation for the Essence software process framework. The body contains no mention of GPL-SLAM, Gaussian processes, laser SLAM, contour representations, recursive Bayesian updates, or associated experiments. Thus, the manuscript as submitted provides an abstract with no supporting technical content.

Significance. If the claims in the abstract were substantiated, a GP-based object-contour SLAM representation with recursive Bayesian updates and confidence bounds could be a useful contribution to laser-based SLAM, particularly for memory-efficient object-level maps and semantic downstream tasks. However, the submitted manuscript does not contain the derivations, algorithms, or experimental results needed to evaluate such a contribution. The significance of the work cannot be assessed from the materials provided; the only concrete, complete content in the full text is a separate study on LLM-based process support, which is outside the claimed scope. There are no machine-checked proofs, reproducible code, or parameter-free derivations to credit in this submission.

major comments (3)
  1. [Abstract vs. Full Text] The abstract claims a novel 'GPL-SLAM' method with GP-based landmarks, recursive contour updates, fully Bayesian joint inference, and validation on synthetic/real-world experiments. The submitted full text is an unrelated paper, 'Using LLMs and Essence to Support Software Practice Adoption' (arXiv:2508.16445v1, cs.SE). No section, equation, table, or figure in the body concerns laser SLAM, Gaussian processes, contour representations, or recursive Bayesian updates. The central claim is therefore entirely unsupported in this manuscript.
  2. [Missing Technical Formulation (Abstract claims)] The claimed GP contour representation, recursive update rule, likelihood model for measurement-to-object association, and joint inference over pose and map are not defined anywhere in the manuscript. Without these, the 'fully Bayesian framework' in the abstract cannot be checked. In particular, the treatment of GP kernel hyperparameters (e.g., length scale and variance) is absent, so how these free parameters are chosen or marginalized is unaddressed.
  3. [Missing Experimental Evidence (Abstract claims)] The abstract says the method is 'validated on synthetic and real world experiments' with 'accurate localization and mapping performance,' but no SLAM experiments, baselines, quantitative metrics (e.g., ATE/RMSE), or uncertainty quantification appear in the submitted text. The tables in §IV report retrieval-augmented generation scores for LLM chatbots, not SLAM performance.
minor comments (3)
  1. [Document metadata] The PDF running head and metadata identify arXiv:2508.16445v1 [cs.SE], while the claimed submission is arXiv:2508.16459 (cs.RO). This identifier mismatch should be resolved in any resubmission.
  2. [Title page] The title and abstract describe a robotics SLAM paper, but the title page and body describe a software-engineering study. The title page and content must match in any corrected submission.
  3. [References] The reference list [1]–[36] supports the LLM/Essence study and has no items on Gaussian processes, SLAM, laser scanning, or object-level mapping; it is unusable for the claimed GPL-SLAM paper.

Circularity Check

0 steps flagged

No circularity found: the submitted full text is an unrelated paper, so the claimed GPL-SLAM derivation chain is absent and cannot reduce to its own inputs.

full rationale

The circularity pass requires exhibiting, by the paper's own equations or by a self-citation chain, that a claimed result is equivalent to its inputs by construction. Here the abstract describes a GPL-SLAM framework with GP-based contour landmarks, recursive online updates, fully Bayesian joint inference, and synthetic/real-world validation, but the full text is an entirely different manuscript (arXiv:2508.16445, 'Using LLMs and Essence to Support Software Practice Adoption') that never mentions GPL-SLAM, Gaussian processes, laser SLAM, contour representations, recursive Bayesian updates, or any relevant experiments. Consequently there is no derivation chain, no fitted parameter renamed as a prediction, no uniqueness theorem imported from prior work, and no ansatz smuggled in via citation that can be quoted and checked. The absence of all supporting content is a severe completeness/integrity problem and makes the abstract's claims unauditable, but it is not an instance of circular reasoning under Rule 1: no specific reduction can be exhibited. Therefore the appropriate circularity score is 0.

Axiom & Free-Parameter Ledger

1 free parameters · 4 axioms · 0 invented entities

The ledger is reconstructed from the abstract only. No new physical entities are introduced; the GP contour landmark is a representation choice, not an invented entity. The central unexamined cost is the GP kernel hyperparameters and the object-decomposability assumption, neither of which is quantified or defended in the abstract.

free parameters (1)
  • GP kernel hyperparameters (e.g., length scale, variance)
    A Gaussian process contour model requires kernel hyperparameters governing smoothness and measurement noise. The abstract does not state whether these are fitted to the experimental data, set by hand, or learned online, so they are an unaudited free parameter of the central claim.
axioms (4)
  • domain assumption The environment is decomposable into distinct objects whose 2D contours are observable by the laser scanner
    Abstract: the map is modeled 'on a per object basis' and validation is in 'diverse structured environments'. If objects cannot be separated, the per-object GP representation collapses.
  • domain assumption Laser measurements can be probabilistically associated with object contours
    Abstract: 'probabilistic measurement to object associations'. Errors in association would corrupt the recursive contour updates and the joint pose-map inference.
  • standard math Standard Gaussian process regression machinery: GP priors, conditioning, posterior updates, chosen kernel
    The GP contour representation relies on standard GP conditioning, which requires Gaussian noise assumptions and a kernel; the abstract invokes 'Gaussian Process (GP) based landmark representations' without proof.
  • standard math Recursive Bayesian filtering for joint pose-map inference
    Abstract: 'fully Bayesian framework, allowing joint inference over the robot pose and object based map'. This is sequential Bayes, assumed rather than derived in the abstract.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of GPL-SLAM: A Laser SLAM Framework with Gaussian Process Based Extended Landmarks." pith.science (2026). https://pith.science/paper/Z4CNF34G

@misc{pith2026250816459,
  author       = {Pith},
  title        = {Pith review of: GPL-SLAM: A Laser SLAM Framework with Gaussian Process Based Extended Landmarks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z4CNF34G}},
  note         = {Machine review of arXiv:2508.16459}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We present a novel Simultaneous Localization and Mapping (SLAM) method that employs Gaussian Process (GP) based landmark (object) representations. Instead of conventional grid maps or point cloud registration, we model the environment on a per object basis using GP based contour representations. These contours are updated online through a recursive scheme, enabling efficient memory usage. The SLAM problem is formulated within a fully Bayesian framework, allowing joint inference over the robot pose and object based map. This representation provides semantic information such as the number of objects and their areas, while also supporting probabilistic measurement to object associations. Furthermore, the GP based contours yield confidence bounds on object shapes, offering valuable information for downstream tasks like safe navigation and exploration. We validate our method on synthetic and real world experiments, and show that it delivers accurate localization and mapping performance across diverse structured environments.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

41 extracted references · 19 canonical work pages · 1 internal anchor

  1. [1]

    The Challenges of Evaluating LLM Applications: An Analysis of Au- tomated, Human, and LLM-Based Approaches

    Bhashithe Abeysinghe and Ruhan Circi. The Challenges of Evaluating LLM Applications: An Analysis of Au- tomated, Human, and LLM-Based Approaches . 2024. arXiv: 2406.03339 ������� . URL: https://arxiv.org/ abs/2406.03339 (cit. on p. 7)

    Pith/arXiv arXiv 2024

  2. [2]

    The AI Scrum Master: Using Large Lan- guage Models (LLMs) to Automate Agile Project Man- agement Tasks

    Zorina Alliata, Tanvi Singhal, and Andreea-Madalina Bozagiu. “The AI Scrum Master: Using Large Lan- guage Models (LLMs) to Automate Agile Project Man- agement Tasks”. In: Agile Processes in Software Engi- neering and Extreme Programming – Workshops . Ed. by Lodovica Marchesi, Alfredo Goldman, Maria Ilaria Lunesu, Adam Przybyłek, Ademar Aguiar, Lorraine Mo...

    work page doi:10.1007/978-3-031-72781-8 2025

  3. [3]

    ChatGPT for Tailoring Software Doc- umentation for Managers and Developers

    Saimir Bala, Kristina Sahling, Jennifer Haase, and Jan Mendling. “ChatGPT for Tailoring Software Doc- umentation for Managers and Developers”. In: Agile Processes in Software Engineering and Extreme Pro- gramming – Workshops. Ed. by Lodovica Marchesi, Al- fredo Goldman, Maria Ilaria Lunesu, Adam Przybyłek, Ademar Aguiar, Lorraine Morgan, Xiaofeng Wang, an...

    doi:10.1007/978- 2025

  4. [4]

    Applications of AI in classical Software Engineering

    Marco Barenkamp, Jonas Rebstadt, and Oliver Thomas. “Applications of AI in classical Software Engineering”. In: AI Perspectives 2 (July 2020). DOI: 10.1186/s42467- 020-00005-4 (cit. on p. 1)

    work page doi:10.1186/s42467- 2020

  5. [5]

    On the Implications of Verbose LLM Outputs: A Case Study in Translation Evaluation

    Eleftheria Briakou, Zhongtao Liu, Colin Cherry, and Markus Freitag. On the Implications of Verbose LLM Outputs: A Case Study in Translation Evaluation. 2024. arXiv: 2410.00863 ������� . URL: https://arxiv.org/ abs/2410.00863 (cit. on p. 6)

    Pith/arXiv arXiv 2024

  6. [6]

    Parizi, and Abbas Yazdinejad

    Talia Crawford, Scott Duong, Richard Fueston, Ay- orinde Lawani, Samuel Owoade, Abel Uzoka, Reza M. Parizi, and Abbas Yazdinejad. AI in Software Engineer- ing: A Survey on Project Management Applications

    work page

  7. [7]

    Towards Effective AI-Powered Agile Project Management

    Hoa Khanh Dam, Truyen Tran, John Grundy, Aditya Ghose, and Yasutaka Kamei. “Towards Effective AI-Powered Agile Project Management”. In: Proc. IEEE/ACM 41st International Conference on Software Engineering: New Ideas and Emerging Results (ICSE- NIER). 2019, pp. 41–44. DOI: 10.1109/ICSE- NIER. 2019.00019 (cit. on p. 1)

    arXiv 2019

  8. [8]

    ACL Ready: RAG Based Assistant for the ACL Checklist

    Michael Galarnyk, Rutwik Routu, Kosha Bheda, Priyan- shu Mehta, Agam Shah, and Sudheer Chava. ACL Ready: RAG Based Assistant for the ACL Checklist

    work page

  9. [9]

    Retrieval-Augmented Generation for Large Language Models: A Survey

    Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, and Haofen Wang. Retrieval-Augmented Generation for Large Language Models: A Survey . 2024. arXiv: 2312. 10997 ������� . URL: https : / / arxiv. org / abs / 2312 . 10997 (cit. on p. 1)

    work page 2024

  10. [10]

    A Fine-Grained Analysis of BERTScore

    Michael Hanna and Ond ˇrej Bojar. “A Fine-Grained Analysis of BERTScore”. In: Proceedings of the Sixth Conference on Machine Translation . Ed. by Loic Bar- rault, Ondrej Bojar, Fethi Bougares, Rajen Chatter- jee, Marta R. Costa-jussa, Christian Federmann, Mark Fishel, Alexander Fraser, Markus Freitag, Yvette Gra- ham, Roman Grundkiewicz, Paco Guzman, Barr...

    work page 2021

  11. [11]

    Large Language Models for Software Engineering: A Systematic Literature Review

    Xinyi Hou, Yanjie Zhao, Yue Liu, Zhou Yang, Kailong Wang, Li Li, Xiapu Luo, David Lo, John Grundy, and Haoyu Wang. “Large Language Models for Software Engineering: A Systematic Literature Review”. In:ACM Trans. Softw. Eng. Methodol. 33.8 (Dec. 2024). ISSN : 1049-331X. DOI: 10.1145/3695988. URL: https://doi. org/10.1145/3695988 (cit. on pp. 1, 2)

    doi:10.1145/3695988 2024

  12. [12]

    The essence of software engineering: the SEMAT kernel

    Ivar Jacobson, Pan-Wei Ng, Paul E. McMahon, Ian Spence, and Svante Lidman. “The essence of software engineering: the SEMAT kernel”. In: Commun. ACM 55.12 (Dec. 2012), pp. 42–49. ISSN : 0001-0782. DOI: 10.1145/2380656.2380670. URL: https://doi.org/10. 1145/2380656.2380670 (cit. on p. 2)

    work page arXiv 2012

  13. [13]

    Enhancing Fairness in LLM Evaluations: Unveiling and Mitigating Biases in Standard-Answer-Based Evaluations

    Tong Jiao, Jian Zhang, Kui Xu, Rui Li, Xi Du, Shangqi Wang, and Zhenbo Song. “Enhancing Fairness in LLM Evaluations: Unveiling and Mitigating Biases in Standard-Answer-Based Evaluations”. In: Proceed- ings of the AAAI Symposium Series 4.1 (Nov. 2024), pp. 56–59. DOI: 10.1609/aaaiss.v4i1.31771. URL: https: //ojs.aaai.org/index.php/AAAI-SS/article/view/3177...

    doi:10.1609/aaaiss.v4i1.31771 2024

  14. [14]

    From LLMs to LLM-based Agents for Software Engineering: A Survey of Cur- rent, Challenges and Future

    Haolin Jin, Linghan Huang, Haipeng Cai, Jun Yan, Bo Li, and Huaming Chen. From LLMs to LLM-based Agents for Software Engineering: A Survey of Cur- rent, Challenges and Future . 2024. arXiv: 2408.02479 ������� . URL: https : / / arxiv. org / abs / 2408 . 02479 (cit. on p. 3)

    Pith/arXiv arXiv 2024

  15. [15]

    Beyond Code Generation: An Observational Study of ChatGPT Usage in Software Engineering Practice

    Ranim Khojah, Mazen Mohamad, Philipp Leitner, and Francisco Gomes de Oliveira Neto. “Beyond Code Generation: An Observational Study of ChatGPT Usage in Software Engineering Practice”. In: Proc. ACM Softw. Eng. 1.FSE (July 2024). DOI: 10.1145/3660788. URL: https://doi.org/10.1145/3660788 (cit. on p. 2)

    doi:10.1145/3660788 2024

  16. [16]

    Software Engineering Education Must Adapt and Evolve for an LLM Environment

    Vassilka D. Kirova, Cyril S. Ku, Joseph R. Laracy, and Thomas J. Marlowe. “Software Engineering Education Must Adapt and Evolve for an LLM Environment”. In: Proceedings of the 55th ACM Technical Symposium on Computer Science Education V . 1. SIGCSE 2024. Port- land, OR, USA: Association for Computing Machinery, 2024, pp. 666–672. DOI: 10.1145/3626252.3630...

    arXiv 2024

  17. [17]

    Retrieval- augmented generation for knowledge-intensive NLP tasks

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K¨uttler, Mike Lewis, Wen-tau Yih, Tim Rockt ¨aschel, Sebastian Riedel, and Douwe Kiela. “Retrieval- augmented generation for knowledge-intensive NLP tasks”. In: Proceedings of the 34th International Confer- ence on Neural Information Processing Systems...

    work page

  18. [18]

    Revolutionizing Retrieval-Augmented Gen- eration with Enhanced PDF Structure Recognition

    Demiao Lin. Revolutionizing Retrieval-Augmented Gen- eration with Enhanced PDF Structure Recognition

    work page

  19. [19]

    SOEN- 101: Code Generation by Emulating Software Process Models Using Large Language Model Agents

    Feng Lin, Dong Jae Kim, and Tse-Hsun Chen. “SOEN- 101: Code Generation by Emulating Software Process Models Using Large Language Model Agents”. In: Proc. IEEE/ACM 47th International Conference on Software Engineering (ICSE). Los Alamitos, CA, USA: IEEE Computer Society, May 2025, pp. 677–677. DOI: 10 . 1109 / ICSE55347 . 2025 . 00140.URL: https : / / doi ...

    work page 2025

  20. [20]

    Can ChatGPT Suggest Patterns? An Exploratory Study About Answers Given by AI- Assisted Tools to Design Problems

    Jo ˜ao Jos ´e Maranh ˜ao Junior, Filipe F. Correia, and Ed- uardo Martins Guerra. “Can ChatGPT Suggest Patterns? An Exploratory Study About Answers Given by AI- Assisted Tools to Design Problems”. In:Agile Processes in Software Engineering and Extreme Programming – Workshops. Ed. by Lodovica Marchesi, Alfredo Gold- man, Maria Ilaria Lunesu, Adam Przybyłek...

    2025

  21. [21]

    A Comprehen- sive Overview of Large Language Models

    Humza Naveed, Asad Ullah Khan, Shi Qiu, Muham- mad Saqib, Saeed Anwar, Muhammad Usman, Naveed Akhtar, Nick Barnes, and Ajmal Mian. A Comprehen- sive Overview of Large Language Models . 2024. arXiv: 2307.06435 ������� . URL: https://arxiv.org/abs/2307. 06435 (cit. on p. 4)

    Pith/arXiv arXiv 2024

  22. [22]

    URL: https://arxiv

    arXiv: 2401.12599 ������� . URL: https://arxiv. org/abs/2401.12599 (cit. on p. 3)

    Pith/arXiv arXiv

  23. [23]

    Improving understanding of the DevOps framework using Essence: a visual rep- resentation

    Paola Nore ˜na-Cardona, Claudia Durango, Elizabeth Suesc´un, and C ´esar Pardo. “Improving understanding of the DevOps framework using Essence: a visual rep- resentation”. In: Periodicals of Engineering and Natural Sciences 13.2 (2025), pp. 305–326. DOI: 10.21533/pen. v13.i2.242 (cit. on p. 3)

    work page doi:10.21533/pen 2025

  24. [24]

    Is Semantic Chunking Worth the Computational Cost?

    Renyi Qu, Ruixuan Tu, and Forrest Sheng Bao. “Is Semantic Chunking Worth the Computational Cost?” In: Findings of the Association for Computational Linguis- tics: NAACL 2025 . Ed. by Luis Chiruzzo, Alan Ritter, and Lu Wang. Albuquerque, New Mexico: Association for Computational Linguistics, Apr. 2025, pp. 2155–

    work page 2025

  25. [25]

    Autonomous Agents in Software Development: A Vi- sion Paper

    Zeeshan Rasheed, Muhammad Waseem, Malik Ab- dul Sami, Kai-Kristian Kemell, Aakash Ahmad, Anh Nguyen Duc, Kari Syst ¨a, and Pekka Abrahamsson. “Autonomous Agents in Software Development: A Vi- sion Paper”. In: Agile Processes in Software Engineer- ing and Extreme Programming – Workshops . Ed. by Lodovica Marchesi, Alfredo Goldman, Maria Ilaria Lunesu, Adam...

    Pith/arXiv arXiv doi:10.48550/arxiv.2311.18440 2025

  26. [26]

    A Comparative Study on Cosine Sim- ilarity Algorithm and Vector Space Model Algorithm on Document Searching

    Warnia Nengsih. “A Comparative Study on Cosine Sim- ilarity Algorithm and Vector Space Model Algorithm on Document Searching”. In: Advanced Science Letters 21.10 (2015), pp. 3321–3323. DOI: 10.1166/asl.2015. 6481 (cit. on p. 4)

    work page doi:10.1166/asl.2015 2015

  27. [27]

    Vendi-RAG: Adaptively Trading-Off Diversity And Quality Significantly Improves Retrieval Augmented Generation With LLMs

    Mohammad Reza Rezaei and Adji Bousso Dieng. Vendi-RAG: Adaptively Trading-Off Diversity And Quality Significantly Improves Retrieval Augmented Generation With LLMs . 2025. arXiv: 2502 . 11228 ������� . URL: https : / / arxiv. org / abs / 2502 . 11228 (cit. on p. 3)

    work page 2025

  28. [28]

    Verbosity Bias in Preference Labeling by Large Language Models

    Keita Saito, Akifumi Wachi, Koki Wataoka, and Youhei Akimoto. Verbosity Bias in Preference Labeling by Large Language Models . 2023. arXiv: 2310 . 10076 ������� . URL: https : / / arxiv. org / abs / 2310 . 10076 (cit. on p. 6)

    work page 2023

  29. [29]

    Optimization Methods for Personalizing Large Lan- guage Models through Retrieval Augmentation

    Alireza Salemi, Surya Kallumadi, and Hamed Zamani. “Optimization Methods for Personalizing Large Lan- guage Models through Retrieval Augmentation”. In: Proceedings of the 47th International ACM SIGIR Con- ference on Research and Development in Information Retrieval. SIGIR ’24. Washington DC, USA: Associa- tion for Computing Machinery, 2024, pp. 752–762. D...

    arXiv 2024

  30. [30]

    Evaluating Re- trieval Quality in Retrieval-Augmented Generation

    Alireza Salemi and Hamed Zamani. “Evaluating Re- trieval Quality in Retrieval-Augmented Generation”. In: Proceedings of the 47th International ACM SIGIR Con- ference on Research and Development in Information Retrieval. SIGIR ’24. Washington DC, USA: Associa- tion for Computing Machinery, 2024, pp. 2395–2400. DOI: 10.1145/3626772.3657957. URL: https://doi...

    arXiv 2024

  31. [31]

    An Empirical Study on Usage and Perceptions of LLMs in a Software Engineering Project

    Sanka Rasnayaka, Guanlin Wang, Ridwan Shariffdeen, and Ganesh Neelakanta Iyer. “An Empirical Study on Usage and Perceptions of LLMs in a Software Engineering Project”. In: Proceedings of the 1st In- ternational Workshop on Large Language Models for Code. LLM4Code ’24. Lisbon, Portugal: Association for Computing Machinery, 2024, pp. 111–118. DOI: 10.1145/3...

    arXiv 2024

  32. [32]

    Diversity Enhances an LLM’s Per- formance in RAG and Long-context Task

    Zhchao Wang, Bin Bi, Yanqi Luo, Sitaram Asur, and Claire Na Cheng. Diversity Enhances an LLM’s Per- formance in RAG and Long-context Task . 2025. arXiv: 2502.09017 ������� . URL: https://arxiv.org/abs/2502. 09017 (cit. on p. 3)

    Pith/arXiv arXiv 2025

  33. [33]

    Evaluation of Retrieval-Augmented Generation: A Survey

    Hao Yu, Aoran Gan, Kai Zhang, Shiwei Tong, Qi Liu, and Zhaofeng Liu. Evaluation of Retrieval-Augmented Generation: A Survey . 2024. arXiv: 2405 . 07437 ������� . URL: https : / / arxiv. org / abs / 2405 . 07437 (cit. on p. 6)

    work page 2024

  34. [34]

    SPRIG: Improving Large Language Model Performance by System Prompt Optimization

    Lechen Zhang, Tolga Ergen, Lajanugen Logeswaran, Moontae Lee, and David Jurgens. SPRIG: Improving Large Language Model Performance by System Prompt Optimization. 2024. arXiv: 2410 . 14826 ������� . URL: https://arxiv.org/abs/2410.14826 (cit. on p. 4)

    Pith/arXiv arXiv 2024

  35. [35]

    Weinberger, and Yoav Artzi

    Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger, and Yoav Artzi. BERTScore: Evaluating Text Generation with BERT . 2020. arXiv: 1904.09675 ������� . URL: https : / / arxiv. org / abs / 1904 . 09675 (cit. on p. 7)

    Pith/arXiv arXiv 2020

  36. [36]

    Improvements to BM25 and Language Models Exam- ined

    Andrew Trotman, Antti Puurula, and Blake Burgess. “Improvements to BM25 and Language Models Exam- ined”. In: Proceedings of the 19th Australasian Doc- ument Computing Symposium . ADCS ’14. Melbourne, VIC, Australia: Association for Computing Machinery, 2014, pp. 58–65. DOI: 10.1145/2682862.2682863. URL: https://doi.org/10.1145/2682862.2682863 (cit. on p. 4)

    arXiv 2014

  37. [41]

    Qiu, and Lili Qiu

    Siyun Zhao, Yuqing Yang, Zilong Wang, Zhiyuan He, Luna K. Qiu, and Lili Qiu. Retrieval Augmented Gen- eration (RAG) and Beyond: A Comprehensive Survey on How to Make your LLMs use External Data More Wisely. 2024. arXiv: 2409.14924 ������� . URL: https: //arxiv.org/abs/2409.14924 (cit. on p. 3)

    Pith/arXiv arXiv 2024

  38. [2020]

    URL: https://arxiv.org/abs/2005.11401 (cit. on p. 3)

    Pith/arXiv arXiv 2005

  39. [2023]

    AI in Software Engineering: A Survey on Project Management Applications

    arXiv: 2307.15224 ������� . URL: https://arxiv. org/abs/2307.15224 (cit. on p. 1)

    work page internal anchor Pith review Pith/arXiv arXiv

  40. [2024]

    URL: https://arxiv

    arXiv: 2408.04675 ������� . URL: https://arxiv. org/abs/2408.04675 (cit. on p. 3)

    work page arXiv

  41. [2177]

    org / 2025

    URL: https : / / aclanthology. org / 2025 . findings - naacl.114/ (cit. on p. 4)

    work page 2025

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.