arxiv: 2604.16353 · v1 · submitted 2026-03-17 · 💻 cs.IR · cs.AI

Recognition: no theorem link

AgriIR: A Scalable Framework for Domain-Specific Knowledge Retrieval

Shuvam Banerji Seal , Aheli Poddar , Alok Mishra , Dwaipayan Roy

Authors on Pith no claims yet

Pith reviewed 2026-05-15 10:41 UTC · model grok-4.3

classification 💻 cs.IR cs.AI

keywords retrieval-augmented generationmodular pipelinedomain-specific retrievalsmall language modelsagricultural informationquery planningtrustworthy retrievalIndian agriculture

0 comments

The pith

AgriIR shows modular stages let 1B-parameter models deliver accurate domain answers without large models.

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

The paper introduces AgriIR as a retrieval-augmented generation framework that splits the answer process into separate declarative stages: query refinement, sub-query planning, retrieval, synthesis, and evaluation. This structure lets users switch to new knowledge areas by changing only the stage details rather than rebuilding the whole system. The reference version applies the approach to Indian agricultural questions by pairing the stages with 1B-parameter language models and adaptive retrievers. Built-in rules force every answer to cite sources and record telemetry so the output stays auditable and reproducible. The central point is that careful pipeline design can produce trustworthy, domain-specific results even when compute resources stay limited.

Core claim

AgriIR decomposes information access into declarative modular stages of query refinement, sub-query planning, retrieval, synthesis, and evaluation. The reference implementation for Indian agriculture combines 1B-parameter language models with adaptive retrievers and domain-aware agent catalogues while enforcing deterministic citation and telemetry collection. This design demonstrates that well-engineered pipelines can achieve domain-accurate, trustworthy retrieval even under constrained resources.

What carries the argument

The declarative modular stages that break the retrieval-augmented generation process into query refinement, sub-query planning, retrieval, synthesis, and evaluation so the system can adapt to new domains without architecture changes.

If this is right

Users can move the same stages to other verticals by editing only the stage logic and domain catalogues.
Small models become viable for specialized retrieval when wrapped in the modular control flow.
Deterministic citation and telemetry make every run auditable and suitable for regulated settings.
Automated deployment assets allow reproducible installs across different hardware environments.

Where Pith is reading between the lines

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

The same stage separation might apply to other low-resource domains such as rural health records or local legal databases.
If the stages prove stable, organizations could maintain domain systems without repeated large-model fine-tuning cycles.
Telemetry data collected at each stage could support later automated improvement of the retrieval components.

Load-bearing premise

The modular stages can transfer to new knowledge verticals without any architecture changes and 1B-parameter models paired with adaptive retrievers will produce accurate grounded answers.

What would settle it

Run the framework on a held-out agricultural query set and measure whether every generated answer matches ground-truth facts and includes correct citations, with failure on more than a small fraction of queries showing the claim does not hold.

Figures

Figures reproduced from arXiv: 2604.16353 by Aheli Poddar, Alok Mishra, Dwaipayan Roy, Shuvam Banerji Seal.

**Figure 1.** Figure 1: AgriIR Configurable Architecture Overview. All components are externally configurable without code modification. tended without altering the overall runtime architecture. The subsequent discussion elaborates the core mechanisms underpinning this workflow, including temperature stratification, parallelization strategy, agentic data curation, adaptive retrieval, and deterministic citation. Stage 1: Query R… view at source ↗

**Figure 2.** Figure 2: Agentic Database Creation Architecture. Autonomous agents (purple) learn from success patterns via persistent tracking (red dashed line), while shared infrastructure (orange) ensures data quality and deduplication across both keyword-based and autonomous approaches. Following are the key architectural features: 1. Persistent Duplicate Tracking: Cross-run deduplication using MD5 content hashing[34] with f… view at source ↗

read the original abstract

This paper introduces AgriIR, a configurable retrieval augmented generation (RAG) framework designed to deliver grounded, domain-specific answers while maintaining flexibility and low computational cost. Instead of relying on large, monolithic models, AgriIR decomposes the information access process into declarative modular stages -- query refinement, sub-query planning, retrieval, synthesis, and evaluation. This design allows practitioners to adapt the framework to new knowledge verticals without modifying the architecture. Our reference implementation targets Indian agricultural information access, integrating 1B-parameter language models with adaptive retrievers and domain-aware agent catalogues. The system enforces deterministic citation, integrates telemetry for transparency, and includes automated deployment assets to ensure auditable, reproducible operation. By emphasizing architectural design and modular control, AgriIR demonstrates that well-engineered pipelines can achieve domain-accurate, trustworthy retrieval even under constrained resources. We argue that this approach exemplifies ``AI for Agriculture'' by promoting accessibility, sustainability, and accountability in retrieval-augmented generation systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

AgriIR describes a modular RAG pipeline for Indian agriculture but supplies no metrics or tests to support its performance claims.

read the letter

The paper's main point is that breaking RAG into fixed stages—query refinement, sub-query planning, retrieval, synthesis, evaluation—plus 1B models and adaptive retrievers can deliver grounded answers in a narrow domain without big compute. That decomposition is the new piece here; it is a straightforward reorganization of existing RAG components tuned to agriculture data and agents, with added emphasis on deterministic citations and telemetry for reproducibility. The architecture description is clear and the deployment assets sound useful for anyone who wants to stand up something similar quickly. The low-parameter choice and modular control are sensible for resource-constrained settings. The central weakness is the complete absence of results. There are no retrieval metrics, no answer accuracy numbers, no ablations, and no comparisons to standard RAG or larger models. The claim that the design achieves domain-accurate, trustworthy retrieval therefore rests on the architecture alone rather than measured outcomes. Readers who already know RAG will see this as a domain-specific template rather than a technical advance. It is worth a serious referee if the authors can add even basic evaluation on agricultural queries; without that it stays a system note. I would bring it to a reading group for the pipeline details but would not cite it until experiments appear.

Referee Report

2 major / 1 minor

Summary. The paper introduces AgriIR, a configurable RAG framework for domain-specific knowledge retrieval that decomposes the process into declarative modular stages (query refinement, sub-query planning, retrieval, synthesis, and evaluation). It targets Indian agriculture using 1B-parameter models, adaptive retrievers, and domain-aware agent catalogues while enforcing deterministic citation, telemetry, and reproducible deployment. The central claim is that this architecture enables domain-accurate, trustworthy, grounded answers under constrained resources and can be adapted to new verticals without architectural changes.

Significance. If the performance claims are validated, the work could be significant for low-resource, domain-specific IR by showing that modular pipelines with small models can deliver accessible and auditable retrieval in agriculture. The emphasis on declarative control, deterministic citation, and deployment assets supports reproducibility and sustainability goals in 'AI for Agriculture'.

major comments (2)

[Abstract] Abstract: The claim that AgriIR 'demonstrates' domain-accurate, trustworthy retrieval is unsupported because the manuscript supplies no quantitative evaluation whatsoever—no retrieval metrics (nDCG, precision@K), no answer accuracy or grounding scores, no ablation studies, and no baseline comparisons (standard RAG or larger models) on agricultural queries.
[No evaluation section present] The manuscript contains no evaluation section or experimental results; the assertion that the declarative stages plus 1B-parameter models produce accurate grounded answers therefore rests entirely on architectural description rather than measured outcomes, which is load-bearing for the central claim.

minor comments (1)

The description of how the five modular stages interact (e.g., how sub-query planning feeds retrieval and how evaluation feeds back) could be clarified with a diagram or pseudocode to improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We agree that the manuscript would be strengthened by the addition of quantitative evaluation and plan to incorporate an evaluation section with the requested metrics, ablations, and baselines in the revised version. The current submission focuses on the declarative architecture and deployment aspects; the performance claims will be supported empirically in the revision.

read point-by-point responses

Referee: [Abstract] Abstract: The claim that AgriIR 'demonstrates' domain-accurate, trustworthy retrieval is unsupported because the manuscript supplies no quantitative evaluation whatsoever—no retrieval metrics (nDCG, precision@K), no answer accuracy or grounding scores, no ablation studies, and no baseline comparisons (standard RAG or larger models) on agricultural queries.

Authors: We agree that the abstract's use of 'demonstrates' is not supported by empirical results in the current manuscript. The paper's primary contribution is the description of a modular, declarative RAG architecture that can be configured for domain-specific use cases with small models. In the revision we will rephrase the abstract to state that AgriIR is designed to enable domain-accurate and trustworthy retrieval under resource constraints, removing any implication of measured performance. We will also add the requested quantitative evaluation section. revision: yes
Referee: [No evaluation section present] The manuscript contains no evaluation section or experimental results; the assertion that the declarative stages plus 1B-parameter models produce accurate grounded answers therefore rests entirely on architectural description rather than measured outcomes, which is load-bearing for the central claim.

Authors: This observation is correct. The submitted manuscript contains no evaluation section and therefore cannot substantiate performance claims with data. We will add a dedicated evaluation section in the revised manuscript that reports retrieval metrics (nDCG, precision@K), answer accuracy and grounding scores, ablation studies on the modular stages, and comparisons against standard RAG pipelines and larger models, all evaluated on Indian agricultural queries. This will directly address the load-bearing nature of the central claim. revision: yes

Circularity Check

0 steps flagged

No circularity: purely descriptive architecture with no derivations or fitted predictions

full rationale

The paper presents a high-level system description of the AgriIR RAG framework, detailing modular stages (query refinement, sub-query planning, retrieval, synthesis, evaluation) and implementation choices such as 1B-parameter models and deterministic citation. No equations, parameter fittings, predictions, or self-citations appear that could reduce any claim to its own inputs by construction. The central assertion that the design achieves domain-accurate retrieval is an untested architectural claim rather than a derived result, so no circular steps exist.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract contains no details on any free parameters, axioms, or new entities; the description is high-level.

pith-pipeline@v0.9.0 · 5475 in / 991 out tokens · 77830 ms · 2026-05-15T10:41:26.157338+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages · 4 internal anchors

[1]

In: Proceedings of the Thirty-Third Text REtrieval Conference (TREC 2024)

Adhikary, S., Banerji Seal, S., Sar, S., Roy, D.: IISERK@ToT_2024: Query re- formulation and layered retrieval for tip-of-tongue items. In: Proceedings of the Thirty-Third Text REtrieval Conference (TREC 2024). National Institute of Stan- dards and Technology (NIST) (2024), https://trec.nist.gov/pubs/trec33/pa pers/IISER-K.tot.pdf

work page 2024
[2]

Almazrouei, E., Alobeidli, H., Alshamsi, A., Cappelli, A., Cojocaru, R., Debbah, M., Étienne Goﬃnet, Hesslow, D., Launay, J., Malartic, Q., Mazzotta, D., Noune, B., Pannier, B., Penedo, G.: The falcon series of open language models (2023), https://arxiv.org/abs/2311.16867

work page arXiv 2023
[3]

In: Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency

Bender, E.M., Gebru, T., McMillan-Major, A., Shmitchell, S.: On the dangers of stochastic parrots: Can language models be too big? . In: Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency. p. 610623. F AccT ’21, Association for Computing Machinery, New York, NY, USA (2021). https://doi.org/10.1145/3442188.3445922 , https:...

work page doi:10.1145/3442188.3445922 2021
[4]

ACM Comput

Bernard, N., Balog, K.: A systematic review of fairness, accountability, trans- parency, and ethics in information retrieval. ACM Comput. Surv. 57(6) (Feb 2025). https://doi.org/10.1145/3637211, https://doi.org/10.1145/3637211

work page doi:10.1145/3637211 2025
[5]

ArXiv (2021), https://crfm.stanford.edu/assets/report.pdf

Bommasani, R., et al., D.A.H.: On the opportunities and risks of foundation mod- els. ArXiv (2021), https://crfm.stanford.edu/assets/report.pdf

work page 2021
[6]

BMC Public Health 25(1), 923 (2025)

Cabrera, M.V., Johnstone, M., Hayward, J., Bolton, K.A., Creighton, D.: Integra- tion of large-scale community-developed causal loop diagrams: a natural language processing approach to merging factors based on semantic similarity. BMC Public Health 25(1), 923 (2025). https://doi.org/10.1186/s12889-025-22142-3 , https://doi.org/10.1186/s12889-025-22142-3

work page doi:10.1186/s12889-025-22142-3 2025
[7]

In: Li, G., Yang, J., Gama, J., Natwichai, J., Tong, Y

Chen, Y., Kuang, J., Cheng, D., Zheng, J., Gao, M., Zhou, A.: Agrikg: An agri- cultural knowledge graph and its applications. In: Li, G., Yang, J., Gama, J., Natwichai, J., Tong, Y. (eds.) Database Systems for Advanced Applications. pp. 533–537. Springer International Publishing, Cham (2019). https://doi.org/10.1 007/978-3-030-18590-9_81 , https://doi.org...

work page doi:10.1007/978-3-030-18590-9_81 2019
[8]

Educational and Psycho- logical Measurement 20(1), 37–46 (1960)

Cohen, J.: A coeﬃcient of agreement for nominal scales. Educational and Psycho- logical Measurement 20(1), 37–46 (1960). https://doi.org/10.1177/00131644 6002000104, https://doi.org/10.1177/001316446002000104

work page doi:10.1177/00131644 1960
[9]

Department of Animal Husbandry and Dairying: Annual report 2022-23. Tech. rep., Ministry of Fisheries, Animal Husbandry and Dairying, Government of India (2023), https://dahd.gov.in/sites/default/files/2023-06/FINALREPORT202 3ENGLISH.pdf

work page 2022
[10]

Douze, M., Guzhva, A., Deng, C., Johnson, J., Szilvasy, G., Mazaré, P.E., Lomeli, M., Hosseini, L., Jégou, H.: The faiss library (2024)

work page 2024
[11]

F AOSTAT Highlights Archive (2025), https://www

Food and Agriculture Organization of the United Nations: Employment indicators 20002023 (july 2025 update). F AOSTAT Highlights Archive (2025), https://www. fao.org/statistics/highlights-archive/highlights-detail/employment-i ndicators-2000-2023-%28july-2025-update%29/en , accessed: 2025-10-27

work page 2025
[12]

In: Melesse, A.M., Abtew, W., Senay, G

Gessesse, A.A., Melesse, A.M.: Chapter 8 - temporal relationships between time series chirps-rainfall estimation and emodis-ndvi satellite images in amhara region, ethiopia. In: Melesse, A.M., Abtew, W., Senay, G. (eds.) Extreme Hydrology and Climate Variability, pp. 81–92. Elsevier (2019). https://doi.org/https://doi. org/10.1016/B978-0-12-815998-9.00008...

work page doi:10.1016/b978-0-12-815998-9.00008-7 2019
[13]

The Llama 3 Herd of Models

Grattaﬁori, A., et al., A.D.: The llama 3 herd of models (2024), https://arxiv. org/abs/2407.21783

work page internal anchor Pith review Pith/arXiv arXiv 2024
[14]

Hugging Face: Massive text embedding benchmark (mteb) leaderboard (2025), https://huggingface.co/spaces/mteb/leaderboard , accessed on 27 October 2025

work page 2025
[15]

Oﬃcial Website (2023), https: //icar.org.in/, accessed: 2025-10-27

ICAR: Indian council of agricultural research. Oﬃcial Website (2023), https: //icar.org.in/, accessed: 2025-10-27

work page 2023
[16]

In: Merlo, P., Tiedemann, J., Tsarfaty, R

Izacard, G., Grave, E.: Leveraging passage retrieval with generative models for open domain question answering. In: Merlo, P., Tiedemann, J., Tsarfaty, R. (eds.) Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume. pp. 874–880. Association for Com- putational Linguistics, Online (Apr 202...

work page doi:10.18653/v1/2021 2021
[17]

ACM Comput

Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y.J., Madotto, A., Fung, P.: Survey of hallucination in natural language generation. ACM Comput. Surv. 55(12) (Mar 2023). https://doi.org/10.1145/3571730 , https://doi.or g/10.1145/3571730 AgriIR: A Scalable Framework for Domain-Speciﬁc Knowledge Retrieval 15

work page doi:10.1145/3571730 2023
[18]

Jiang, A.Q., Sablayrolles, A., Mensch, A., Bamford, C., Chaplot, D.S., de las Casas, D., Bressand, F., Lengyel, G., Lample, G., Saulnier, L., Lavaud, L.R., Lachaux, M.A., Stock, P., Scao, T.L., Lavril, T., Wang, T., Lacroix, T., Sayed, W.E.: Mistral 7b (2023), https://arxiv.org/abs/2310.06825

work page internal anchor Pith review Pith/arXiv arXiv 2023
[19]

IEEE Transactions on Big Data 7(3), 535–547 (2019)

Johnson, J., Douze, M., Jégou, H.: Billion-scale similarity search with GPUs. IEEE Transactions on Big Data 7(3), 535–547 (2019)

work page 2019
[20]

https://doi.org/https://doi.org/10.1016/j.inffus.2025

Katharria, A., Pant, M., Velásquez, J.D., Snáel, V., Rajwar, K., Deep, K.: Informa- tion fusion in smart agriculture: machine learning applications and future research directions (2026). https://doi.org/https://doi.org/10.1016/j.inffus.2025. 104040, https://www.sciencedirect.com/science/article/pii/S15662535250 11029

work page doi:10.1016/j.inffus.2025 2026
[21]

International Journal on Digital Libraries 25(4), 569584 (Jun 2023)

Koopman, B., Mourad, A., Li, H., Vegt, A.v.d., Zhuang, S., Gibson, S., Dang, Y., Lawrence, D., Zuccon, G.: Agask: an agent to help answer farmers questions from scientiﬁc documents. International Journal on Digital Libraries 25(4), 569584 (Jun 2023). https://doi.org/10.1007/s00799-023-00369-y , http://dx.doi.org/1 0.1007/s00799-023-00369-y

work page doi:10.1007/s00799-023-00369-y 2023
[22]

On the interpretation of weight vectors of linear models in multivariate neuroimaging.Neuroimage, 87:96–110, 2014

Kuska, M.T., Wahabzada, M., Paulus, S.: Ai for crop production where can large language models (llms) provide substantial value? Computers and Electronics in Agriculture 221, 108924 (2024). https://doi.org/https://doi.org/10.1016/j. compag.2024.108924, https://www.sciencedirect.com/science/article/pii/ S0168169924003156

work page doi:10.1016/j 2024
[23]

In: Larochelle, H., Ran- zato, M., Hadsell, R., Balcan, M., Lin, H

Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küt- tler, H., Lewis, M., Yih, W.t., Rocktäschel, T., Riedel, S., Kiela, D.: Retrieval- augmented generation for knowledge-intensive nlp tasks. In: Larochelle, H., Ran- zato, M., Hadsell, R., Balcan, M., Lin, H. (eds.) Advances in Neural Informa- tion Processing Systems. vol. 33, pp....

work page 2020
[24]

Lin, X., Ning, Y., Zhang, J., Dong, Y., Liu, Y., Wu, Y., Qi, X., Sun, N., Shang, Y., Cao, P., Zou, L., Chen, X., Zhou, C., Wu, J., Pan, S., Wang, B., Cao, Y., Chen, K., Hu, S., Guo, L.: Llm-based agents suﬀer from hallucinations: A survey of taxonomy, methods, and directions (2025), https://arxiv.org/abs/2509.18970

work page arXiv 2025
[25]

In: Proceedings of the 30th ACM International Conference on Information & Knowledge Management

Macdonald, C., Tonellotto, N., MacA vaney, S., Ounis, I.: Pyterrier: Declarative experimentation in python from bm25 to dense retrieval. In: Proceedings of the 30th ACM International Conference on Information & Knowledge Management. p. 45264533. CIKM ’21, Association for Computing Machinery, New York, NY, USA (2021). https://doi.org/10.1145/3459637.348201...

work page doi:10.1145/3459637.3482013 2021
[26]

ACM Trans

Mai, G., Huang, W., Sun, J., Song, S., Mishra, D., Liu, N., Gao, S., Liu, T., Cong, G., Hu, Y., Cundy, C., Li, Z., Zhu, R., Lao, N.: On the opportunities and challenges of foundation models for geoai (vision paper). ACM Trans. Spatial Algorithms Syst. 10(2) (Jul 2024). https://doi.org/10.1145/3653070, https://doi.org/10.114 5/3653070

work page doi:10.1145/3653070 2024
[27]

Maliakel, P.J., Ilager, S., Brandic, I.: Investigating energy eﬃciency and perfor- mance trade-oﬀs in llm inference across tasks and dvfs settings (2025), https: //arxiv.org/abs/2501.08219

work page arXiv 2025
[28]

In: Proceedings of the Conference on Fairness, Accountability, and Transparency

Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., Spitzer, E., Raji, I.D., Gebru, T.: Model cards for model reporting. In: Proceedings of the Conference on Fairness, Accountability, and Transparency. p. 220229. F AT* ’19, Association for Computing Machinery, New York, NY, USA (2019). https: 16 Banerji Seal et al. //doi.org/10....

work page doi:10.1145/3287560.3287596 2019
[29]

Mustofa, S., Munna, M.M.H., Emon, Y.R., Rabbany, G., Ahad, M.T.: A com- prehensive review on plant leaf disease detection using deep learning (2023), https://arxiv.org/abs/2308.14087

work page arXiv 2023
[30]

SIGIR Forum 53(2), 2043 (Mar 2021)

Olteanu, A., Garcia-Gathright, J., de Rijke, M., Ekstrand, M.D., Roegiest, A., Li- pani, A., Beutel, A., Olteanu, A., Lucic, A., Stoica, A.A., Das, A., Biega, A., Voorn, B., Hauﬀ, C., Spina, D., Lewis, D., Oard, D.W., Yilmaz, E., Hasibi, F., Kazai, G., McDonald, G., Haned, H., Ounis, I., van der Linden, I., Garcia-Gathright, J., Baan, J., Lau, K.N., Balog...

work page doi:10.1145/3458553.3458556 2043
[31]

Peng, R., Liu, K., Yang, P., Yuan, Z., Li, S.: Embedding-based retrieval with llm for eﬀective agriculture information extracting from unstructured data (2023), https://arxiv.org/abs/2308.03107

work page arXiv 2023
[32]

Sentence- BERT : Sentence Embeddings using S iamese BERT -Networks

Reimers, N., Gurevych, I.: Sentence-bert: Sentence embeddings using siamese bert- networks. In: Inui, K., Jiang, J., Ng, V., Wan, X. (eds.) Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th In- ternational Joint Conference on Natural Language Processing (EMNLP-IJCNLP). pp. 3982–3992. Association for Computa...

work page doi:10.18653/v1/d19-1410 2019
[33]

In: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

Reimers, N., Gurevych, I.: Making monolingual sentence embeddings multilingual using knowledge distillation. In: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). pp. 4512–4525. Association for Computational Linguistics, Online (Nov 2020). https://doi.org/10.18653/v 1/2020.emnlp-main.365, https://aclanthology....

work page doi:10.18653/v 2020
[34]

Rivest, R.: The md5 message-digest algorithm (RFC1321) (1992), http://www.ie tf.org/rfc/rfc1321.txt

work page 1992
[35]

In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

Salvador, A., Hynes, N., Aytar, Y., Marin, J., Oﬂi, F., Weber, I., Torralba, A.: Learning cross-modal embeddings for cooking recipes and food images. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE (Jul 2017)

work page 2017
[36]

Samuel, D.J., Skarga-Bandurova, I., Sikolia, D., Awais, M.: Agrollm: Connecting farmers and agricultural practices through large language models for enhanced knowledge transfer and practical application (2025), https://arxiv.org/abs/25 03.04788

work page 2025
[37]

Progress in Ar- tiﬁcial Intelligence 14(2), 117–164 (2025)

Shaikh, T.A., Rasool, T., Veningston, K., Yaseen, S.M.: The role of large language models in agriculture: harvesting the future with llm intelligence. Progress in Ar- tiﬁcial Intelligence 14(2), 117–164 (2025). https://doi.org/10.1007/s13748-0 24-00359-4 , https://doi.org/10.1007/s13748-024-00359-4

work page doi:10.1007/s13748-0 2025
[38]

In: Korhonen, A., Traum, D., Màrquez, L

Strubell, E., Ganesh, A., McCallum, A.: Energy and policy considerations for deep learning in NLP. In: Korhonen, A., Traum, D., Màrquez, L. (eds.) Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. pp. 3645–3650. Association for Computational Linguistics, Florence, Italy (Jul 2019). https://doi.org/10.18653/v1/P19-135...

work page internal anchor Pith review doi:10.18653/v1/p19-1355 2019
[39]

Team, G.: Gemma 3 technical report (2025), https://arxiv.org/abs/2503.19786 AgriIR: A Scalable Framework for Domain-Speciﬁc Knowledge Retrieval 17

work page internal anchor Pith review Pith/arXiv arXiv 2025
[40]

Team, O.: Ollama: An open source framework for running and serving large lan- guage models locally (2023), https://github.com/ollama/ollama, version latest, accessed 27 October 2025

work page 2023
[41]

In: Thirty-ﬁfth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2) (2021), https://openreview.net/forum?id=wCu6 T5xFjeJ

Thakur, N., Reimers, N., Rücklé, A., Srivastava, A., Gurevych, I.: BEIR: A het- erogeneous benchmark for zero-shot evaluation of information retrieval models. In: Thirty-ﬁfth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2) (2021), https://openreview.net/forum?id=wCu6 T5xFjeJ

work page 2021
[42]

Artiﬁcial Intelligence Review 58(3), 92 (2025)

Upadhyay, A., Chandel, N.S., Singh, K.P., Chakraborty, S.K., Nandede, B.M., Kumar, M., Subeesh, A., Upendar, K., Salem, A., Elbeltagi, A.: Deep learning and computer vision in plant disease detection: a comprehensive review of techniques, models, and trends in precision agriculture. Artiﬁcial Intelligence Review 58(3), 92 (2025). https://doi.org/10.1007/s...

work page doi:10.1007/s10462-024-11100-x 2025
[43]

In: Proceedings of the Australasian Conference on Informa- tion Systems (ACIS 2024)

Wilson, S., Ginige, A., Goonatilake, J.: Design science research approach for ontology development in agriculture: Utilising advances of llm for automated entity extraction. In: Proceedings of the Australasian Conference on Informa- tion Systems (ACIS 2024). No. 150, Association for Information Systems (2024), https://aisel.aisnet.org/acis2024/150, aCIS 2...

work page 2024
[44]

https://doi.org/10.1007/978-981-96-0 573-6_21, https://doi.org/10.1007/978-981-96-0573-6_21

Yang, S., Liu, Z., Mayer, W., Ding, N., Wang, Y., Huang, Y., Wu, P., Li, W., Li, L., Zhang, H.Y., Feng, Z.: Shizishangpt: An agricultural large language model integrating tools and resources (2024). https://doi.org/10.1007/978-981-96-0 573-6_21, https://doi.org/10.1007/978-981-96-0573-6_21

work page doi:10.1007/978-981-96-0 2024