Pith. sign in

REVIEW 3 major objections 4 minor 46 references

Leveraging Knowledge Graphs and LLM Reasoning to Identify Operational Bottlenecks for Warehouse Planning Assistance

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Turning DES warehouse logs into a knowledge graph, an LLM agent that reasons in self-corrected steps answers operational questions near-perfectly and diagnoses planted bottlenecks more precisely than baseline methods.

desk verdict A plausible KG+LLM feasibility study with a small but real operational QA win; the headline diagnostic-ability claim is not yet backed by scored evidence. read the letter →

arxiv 2507.17273 v1 pith:5NJNDZA3 submitted 2025-07-23 cs.LG cs.AI

Rishi Parekh , Saisubramaniam Gopalakrishnan , Zishan Ahmad , Anirudh Deodhar This is my paper
classification cs.LGcs.AI
keywords discreteeventsimulationknowledgegraphLLMagentsCypherquerygenerationbottleneckdiagnosiswarehouseoperationsself-reflectioniterativereasoning
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

Warehouse planners who run discrete event simulations are left with huge event logs that take manual effort or custom scripts to interpret. This paper tries to make that interpretation conversational and automatic: it converts the simulation output into a knowledge graph whose edges are package movements and whose nodes are suppliers, workers, AGVs, forklifts, and storage blocks, then lets an LLM-based agent reason over the graph. The agent decomposes each question into sub-questions, translates each into graph queries, executes them, and self-corrects, so that it can both answer factual operational questions and trace root causes of planted bottlenecks. The authors demonstrate on one unloading-warehouse simulation that their step-wise approach answers 25 operational questions with near-perfect pass rates and, in three case studies, gives diagnoses that align with a human expert while being more specific than baseline methods. If the approach generalizes, it would cut the time from simulation run to actionable insight and make root-cause analysis repeatable for warehouse planning.

What carries the argument

The load-bearing object is a warehouse knowledge graph built from DES logs: five node types (supplier, worker, AGV, forklift, storage) with timestamps on nodes and edges, where each package's journey from supplier to storage is a path of typed relationships. The reasoning mechanism is a dual-path LLM agent. Operational questions go through a QA chain that first generates a step-wise plan, then for each step writes a Cypher query (the graph query language), executes it, and self-reflects before assembling the answer. Investigative questions go through an iterative reasoning chain that formulates one sub-question at a time, runs each through the full QA chain, and conditions the next sub-question on the accumulated evidence until a summarizer produces the root-cause narrative. The design choice of Cypher over SQL is meant to let the agent express multi-hop operational patterns directly on the graph structure.

What would settle it

Take the same framework to an independently built DES warehouse model with a different layout and process flow, have a separate team write the operational questions and plant one bottleneck without telling the agent's designers, then check whether the agent's root-cause summary names the planted bottleneck and its stage. If its diagnostic accuracy or pass rates drop substantially compared with the paper's 0.82 Pass@1 and perfect Pass@4, the reported performance is tied to this specific testbed rather than to the method.

Watch

Extended reading notes

Core claim

The central claim is that a KG+LLM agent can serve as the analysis layer for DES output: structured event data becomes a queryable graph, and an LLM that plans sub-questions, writes Cypher, and reflects on its own answers can reliably identify where warehouse operations are failing. For the 25 operational questions, the Step-wise Guide achieves an average Pass@1 of 0.82 and Pass@4 of 1.00, compared with 0.41/0.56 for single-pass Direct QA and 0.73/0.80 for Direct QA with self-reflection. For the investigative scenarios, the agent pinpoints the specific failing stage and entity, for instance isolating the AGV-to-forklift transfer for supplier CamelCargo's delay, supplier AuroraFarms' processing for the 10:00-12:30 slowdown, and forklift FL_00's waiting and movement times in the third scenario. In each case the authors report that the agent's summary matched the human expert's conclusion while adding quantitative support and ruling out alternative causes that the baselines missed.

Load-bearing premise

The claim rests on the assumption that one simulated unloading warehouse, the 25 author-written questions with their gold answers, and the three planted bottleneck scenarios are representative enough that the measured pass rates and diagnostic quality will carry over to real warehouse planning data.

Editorial extensions

If this is right

  • On the paper's 25-question benchmark, every operational question is answered correctly within four attempts, so planners could rely on the agent for routine status checks and KPI-style questions.
  • Root-cause questions no longer require inspect-the-log detective work: the agent's summaries identify the specific stage, supplier, or equipment causing a delay, with numbers attached.
  • The per-step self-reflection mechanism catches query errors before they propagate, which is why decomposed step-wise answering beats single-shot query generation on multi-fact questions.
  • The same KG+LLM pipeline turns a warehouse digital twin from a passive simulator into an interactive assistant that can be asked about variability, utilization, and bottlenecks in plain language.

Reading between the lines

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

  • If the method transfers, the recipe is portable to any DES whose output is event logs with entity IDs and timestamps: manufacturing lines, terminal operations, or hospital patient flow could use the same graph-schema-and-agent pattern.
  • The biggest near-term gain is probably in re-runs: because the KG construction is automated, the same assistant can be pointed at each new simulation scenario and immediately produce comparable diagnostic narratives.
  • A testable extension would be to feed the agent logs with two simultaneously planted bottlenecks and see whether the iterative chain surfaces both; the current scenarios each introduce one primary inefficiency.
  • Since Pass@4 reaches 1.00 while Pass@1 is 0.82, most remaining errors may be isolated query slips rather than reasoning failures, suggesting that running the step chain a few times and aggregating answers could close most of the gap.
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

3 major / 4 minor

Summary. The paper proposes a framework that converts raw Discrete Event Simulation (DES) output from a warehouse unloading process into a Knowledge Graph (KG) and uses an LLM-based agent with iterative reasoning (sub-questioning, Cypher generation, self-reflection) to answer operational questions and diagnose planted bottlenecks. The evaluation consists of a quantitative Pass@k comparison on a set of 25 (or per Table 6, 26) author-authored operational questions against two baselines, plus three qualitative case studies for investigative bottleneck diagnosis.

Significance. If validated, the framework could meaningfully reduce time-to-insight for DES analysis and make simulation outputs queryable by warehouse planners without specialized query languages. The paper provides a clear system architecture, a concrete KG schema, and a transparent—though small—evaluation of operational question answering. The inclusion of the full question list and the description of the warehouse simulation are helpful for replication, but no code, data, or exact executed Cypher queries are released. The central differentiator claimed by the paper—superior diagnostic ability for investigative questions—is not established by the current evidence, because it rests on unblinded, qualitative case studies.

major comments (3)
  1. [Section 5.2 and Appendix A.3] The claim that the framework demonstrates "superior diagnostic ability" is based on three case studies in which the human expert already knew that a bottleneck had been planted and where the comparison to the baseline is entirely qualitative (e.g., "general", "broad", "limited further diagnosis") with no scoring rubric, no predefined success criteria, and no independent gold standard. No quantitative measure is reported for whether the agent identified the planted bottleneck, how many correct causal factors it found, or how many false attributions it made. Consequently, the paper's main differentiator over existing methods is not established by the reported evidence.
  2. [Table 1 and Section 3.2.1] The operational QA claim rests on an author-built set of 25 (or, per Table 6, 26) questions with roughly five per category. No confidence intervals or significance tests are provided, so the reported differences (e.g., Direct QA+SR Pass@1 0.73 vs. Step-wise Guide 0.82) may be within sampling noise. Moreover, the abstract's "near-perfect pass rates" refer to Pass@4=1.00, not Pass@1=0.82; Pass@4 measures success over four attempts and is substantially weaker than first-attempt accuracy. The authors should report Pass@1 with error bars and avoid conflating the two metrics.
  3. [Tables 2, 3, 5 and Appendix A.3] The case-study evidence is not reproducible from the paper: the displayed Cypher queries are "highly condensed conceptual representations" and the executed queries and result sets are not provided. The reported KG results also contain implausible entries—for instance, Table 5 lists "Util 0" for every forklift, which is inconsistent with the text's discussion of utilization rates and suggests that the tables are illustrative rather than exact. This makes it impossible to verify the factual basis of the diagnostic conclusions, so the case studies should be treated as illustrative examples rather than validated comparative results.
minor comments (4)
  1. [Section 3.2.1 and Table 6] Section 3.2.1 states that there are 25 operational questions, but Table 6 contains 26 entries; the counts and per-category breakdown should be reconciled.
  2. [Abstract] The phrase "near-perfect pass rates in pinpointing inefficiencies" should be qualified as Pass@4 performance, with the Pass@1 value (0.82) reported in the same sentence for clarity.
  3. [Section 5.1] The phrase "maximum Pass@4 scores across all operational stages" is ambiguous; stating directly that all Step-wise Guide Pass@4 values are 1.00 would be clearer.
  4. [Table 5] The "Util 0" entries for every forklift in the first row of Table 5 should be corrected or the query formula that produced them should be explained; as written, they contradict the utilization definitions used elsewhere in the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the framework is evaluated against fixed gold answers and planted bottleneck scenarios, not fitted to its own outputs.

full rationale

The paper's derivation chain is DES output -> KG construction -> LLM agent query generation -> answer synthesis. No parameter is fitted to the evaluation data and no predicted quantity is defined in terms of the target. Operational QA performance (Table 1) is measured with Pass@k against a fixed list of 25 author-authored questions (Table 6) with gold answers; although the benchmark is small and self-created, correctness is an external match to those gold answers rather than a consequence of the framework's design. The investigative claims rest on three planted-bottleneck scenarios; the agent's narrative diagnoses are compared with a human expert who knew the planted bottleneck, which weakens evaluator blinding but does not make the agent's output equivalent to its input. The one self-citation (reference [29], a co-authored WSDM paper on hallucination mitigation) appears only as background for RAG in Related Work and is not load-bearing for any central argument. The Appendix note that Cypher queries are 'highly condensed conceptual representations' is a reproducibility limitation, not a circular step. Consequently, no step in the claimed derivation reduces to its own input.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claim depends on the representativeness of the simulated scenario, the completeness of the KG schema, the reliability of LLM-generated Cypher queries, and the validity of the author-built evaluation. No parameters are fitted to data; the hand-chosen settings above are part of the experimental protocol and constrain generalization.

free parameters (3)
  • Pass@4 evaluation horizon (k=4) = 4 attempts
    The headline near-perfect pass rates are Pass@4 scores, not single-attempt Pass@1 (which is 0.82 average). This choice is not fit to data, but it is a hand-picked evaluation protocol that directly shapes the central claim.
  • LLM inference settings (temperature, top_p, token limit) = temperature 0.0, top_p 0.95, 4096 tokens
    These fixed sampling settings, stated in Section 5, can affect Cypher generation and answer quality; the reported results are conditional on them.
  • Warehouse scenario configuration = 5 suppliers, 12 workers, 20 AGVs, 5 forklifts, 5 blocks, 30-minute arrival intervals, 60-90 second storage times
    These hand-chosen simulation parameters define the only dataset on which the framework was evaluated; the central empirical claims are not demonstrated for other warehouse configurations.
assumptions (6)
  • domain assumption The simulated unloading scenario is representative of real warehouse operations, and the injected bottlenecks are detectable in the extracted logs.
    Section 3.1 configures a single warehouse unloading model; Section 5.2 uses three planted bottleneck scenarios to judge diagnostic ability.
  • domain assumption The KG schema preserves all information needed for bottleneck diagnosis; no relevant temporal or relational data is lost when DES logs are converted to nodes and edges.
    Section 4.1 and Appendix A.2 define the schema; if an edge or timestamp is missing, Cypher queries cannot recover it.
  • domain assumption GPT-4o can generate valid, semantically correct Cypher queries for the schema, and the self-reflection loop catches a meaningful fraction of errors.
    Section 5 states interaction uses OpenAI GPT-4o; Section 4.2 relies on self-reflection for correction; the paper's own limitations (Section 6) concede reliability needs ongoing evaluation.
  • domain assumption The human expert's qualitative analysis provides correct ground truth for the investigative scenarios.
    Section 5.2 and Appendix A.3 compare agent summaries to a human expert; if the expert analysis is wrong or incomplete, the claimed diagnostic superiority is unsupported.
  • domain assumption The 25 operational questions and their gold answers are correct and unambiguous.
    Section 3.2.1 and Appendix Table 6 present the questions, but the paper does not show how gold answers were generated or independently verified.
  • domain assumption Pass@k with k=4 is an adequate measure of answer accuracy for the near-perfect claim.
    Section 5.1 reports Pass@1 and Pass@4; the abstract highlights near-perfect rates, which only hold at Pass@4.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Leveraging Knowledge Graphs and LLM Reasoning to Identify Operational Bottlenecks for Warehouse Planning Assistance." pith.science (2026). https://pith.science/paper/5NJNDZA3

@misc{pith2026250717273,
  author       = {Pith},
  title        = {Pith review of: Leveraging Knowledge Graphs and LLM Reasoning to Identify Operational Bottlenecks for Warehouse Planning Assistance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5NJNDZA3}},
  note         = {Machine review of arXiv:2507.17273}
}
read the original abstract

Analyzing large, complex output datasets from Discrete Event Simulations (DES) of warehouse operations to identify bottlenecks and inefficiencies is a critical yet challenging task, often demanding significant manual effort or specialized analytical tools. Our framework integrates Knowledge Graphs (KGs) and Large Language Model (LLM)-based agents to analyze complex Discrete Event Simulation (DES) output data from warehouse operations. It transforms raw DES data into a semantically rich KG, capturing relationships between simulation events and entities. An LLM-based agent uses iterative reasoning, generating interdependent sub-questions. For each sub-question, it creates Cypher queries for KG interaction, extracts information, and self-reflects to correct errors. This adaptive, iterative, and self-correcting process identifies operational issues mimicking human analysis. Our DES approach for warehouse bottleneck identification, tested with equipment breakdowns and process irregularities, outperforms baseline methods. For operational questions, it achieves near-perfect pass rates in pinpointing inefficiencies. For complex investigative questions, we demonstrate its superior diagnostic ability to uncover subtle, interconnected issues. This work bridges simulation modeling and AI (KG+LLM), offering a more intuitive method for actionable insights, reducing time-to-insight, and enabling automated warehouse inefficiency evaluation and diagnosis.

Figures

Figures reproduced from arXiv: 2507.17273 by the authors.

Figure 1
Figure 1. Workflow diagram of the warehouse unload [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. System architecture of the LLM Reasoning Agent [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

46 extracted references · 35 canonical work pages

  1. [29]

    I Muneeswaran, Advaith Shankar, V Varun, Saisubramaniam Gopalakrish- nan, and Vishal Vaddina. 2024. Mitigating Factual Inconsistency and Hallu- cination in Large Language Models.. In WSDM. 1169–1170

    work page 2024

  2. [1]

    Bilal Abu-Salih. 2021. Domain-specific knowledge graphs: A survey. Journal of Network and Computer Applications 185 (2021), 103076

    work page 2021

  3. [2]

    Agalianos, S.T

    K. Agalianos, S.T. Ponis, E. Aretoulaki, G. Plakas, and O. Efthymiou. 2020. Discrete Event Simulation and Digital Twins: Review and Challenges for Logistics. Procedia Manufacturing 51 (2020), 1636–1641. doi:10.1016/j.promfg. 2020.10.228 30th International Conference on Flexible Automation and Intelligent Manufacturing (FAIM2021)

    work page doi:10.1016/j.promfg 2020

  4. [3]

    Garima Agrawal, Tharindu Kumarage, Zeyad Alghamdi, and Huan Liu. 2023. Can knowledge graphs reduce hallucinations in llms?: A survey. arXiv preprint arXiv:2311.07914 (2023)

    arXiv 2023

  5. [4]

    Marco Arazzi, Davide Ligari, Serena Nicolazzo, and Antonino Nocera. 2025. Augmented Knowledge Graph Querying leveraging LLMs. arXiv preprint arXiv:2502.01298 (2025)

    work page arXiv 2025

  6. [5]

    Jerry Banks. 2005. Discrete event system simulation . Pearson Education

    work page 2005

  7. [6]

    Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde De Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. 2021. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374 (2021)

    arXiv 2021

  8. [7]

    René De Koster, Tho Le-Duc, and Kees Jan Roodbergen. 2007. Design and control of warehouse order picking: A literature review. European journal of operational research 182, 2 (2007), 481–501

    work page 2007

Show all 46 references
  1. [8]

    Adnane Drissi Elbouzidi, Abdessamad Ait El Cadi, Robert Pellerin, Samir Lamouri, Estefania Tobon Valencia, and Marie-Jane Bélanger. 2023. The Role of AI in Warehouse Digital Twins: Literature Review. Applied Sciences 13, 11 (2023). https://www.mdpi.com/2076-3417/13/11/6746

    2023

  2. [9]

    Jinxiang Gu, Marc Goetschalckx, and Leon F McGinnis. 2007. Research on warehouse operation: A comprehensive review. European journal of operational research 177, 1 (2007), 1–21

    2007

  3. [10]

    Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V Chawla, Olaf Wiest, and Xiangliang Zhang. 2024. Large language model based multi-agents: A survey of progress and challenges. arXiv preprint arXiv:2402.01680 (2024)

    2024 arXiv

  4. [11]

    Domen Hočevar and Klemen Kenda. 2024. Integrating Knowledge Graphs and Large Language Models for Querying in an Industrial Environment . Ph. D. Dissertation. Bachelor’s Thesis. University of Ljubljana

    2024

  5. [12]

    Aidan Hogan, Eva Blomqvist, Michael Cochez, Claudia d’Amato, Gerard De Melo, Claudio Gutierrez, Sabrina Kirrane, José Emilio Labra Gayo, Roberto Navigli, Sebastian Neumaier, et al. 2021. Knowledge graphs. ACM Computing Surveys (Csur) 54, 4 (2021), 1–37

    2021

  6. [13]

    Markus Hornsteiner, Michael Kreussel, Christoph Steindl, Fabian Ebner, Philip Empl, and Stefan Schönig. 2024. Real-Time Text-to-Cypher Query Generation with Large Language Models for Graph Databases. Future Inter- net 16, 12 (2024), 438

    2024

  7. [14]

    Jiaxin Huang, Shixiang Shane Gu, Le Hou, Yuexin Wu, Xuezhi Wang, Hongkun Yu, and Jiawei Han. 2022. Large language models can self-improve. arXiv preprint arXiv:2210.11610 (2022)

    2022 arXiv

  8. [15]

    Saverio Ieva, Ivano Bilenchi, Filippo Gramegna, Agnese Pinto, Floriano Scioscia, Michele Ruta, and Giuseppe Loseto. 2025. Enhancing Last-Mile Logistics: AI-Driven Fleet Optimization, Mixed Reality, and Large Language Model Assistants for Warehouse Operations. Sensors 25, 9 (20...

    2025

  9. [16]

    Dmitry Ivanov, Alexandre Dolgui, and Boris Sokolov. 2019. The impact of digital technology and Industry 4.0 on the ripple effect and supply chain risk analytics. International journal of production research 57, 3 (2019), 829–846

    2019

  10. [17]

    Jinhao Jiang, Kun Zhou, Wayne Xin Zhao, Yang Song, Chen Zhu, Hengshu Zhu, and Ji-Rong Wen. 2024. Kg-agent: An efficient autonomous agent framework for complex reasoning over knowledge graph. arXiv preprint arXiv:2402.11163 (2024)

    2024 arXiv

  11. [18]

    Ajay Kattepur and Balamuralidhar P. 2019. Roboplanner: autonomous robotic action planning via knowledge graph queries. In Proceedings of the 34th ACM/SIGAPP Symposium on Applied Computing . 953–956

    2019

  12. [19]

    Edward Elson Kosasih, Fabrizio Margaroli, Simone Gelli, Ajmal Aziz, Nick Wildgoose, and Alexandra Brintrup. 2024. Towards knowledge graph rea- soning for supply chain risk management using graph neural networks. International Journal of Production Research 62, 15 (2024), 5596–5612

    2024

  13. [20]

    Averill M Law, W David Kelton, and W David Kelton. 2000. Simulation modeling and analysis. Vol. 3. Mcgraw-hill New York

    2000

  14. [21]

    Carman KM Lee, Yaqiong Lv, Kam KH Ng, William Ho, and King Lun Choy

  15. [22]

    Jiewu Leng, Hao Zhang, Douxi Yan, Qiang Liu, Xin Chen, and Ding Zhang

  16. [23]

    Donghe Li, Zuchen Li, Ye Yang, Li Sun, Dou An, and Qingyu Yang. 2024. Knowledge graph-enhanced large language model for domain-specific ques- tion answering systems. Authorea Preprints (2024)

    2024

  17. [24]

    Linhao Luo, Yuan-Fang Li, Gholamreza Haffari, and Shirui Pan. 2023. Rea- soning on graphs: Faithful and interpretable large language model reasoning. arXiv preprint arXiv:2310.01061 (2023)

    2023 arXiv

  18. [25]

    Aman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan, Luyu Gao, Sarah Wiegreffe, Uri Alon, Nouha Dziri, Shrimai Prabhumoye, Yiming Yang, et al. 2023. Self-refine: Iterative refinement with self-feedback. Advances in Neural Information Processing Systems 36 (2023), 46534–46594

    2023

  19. [26]

    Ioanna Mandilara, Christina Maria Androna, Eleni Fotopoulou, Anastasios Zafeiropoulos, and Symeon Papavassiliou. 2025. Decoding the Mystery: How can LLMs Turn Text into Cypher in Complex Knowledge Graphs?IEEE Access (2025)

    2025

  20. [27]

    Hokey Min. 2010. Artificial intelligence in supply chain management: theory and applications. International Journal of Logistics: Research and Applications 13, 1 (2010), 13–39

    2010

  21. [28]

    Sebastian Monkaa, Irlan Grangel-Gonzáleza, Stefan Schmida, Lavdim Halilaja, Marc Rickartb, Oliver Rudolphb, and Rui Diasb. [n. d.]. Enhancing Manufacturing Knowledge Access with LLMs and Context-aware Prompting. ([n. d.])

  22. [30]

    Natasha Noy, Yuqing Gao, Anshu Jain, Anant Narayanan, Alan Patterson, and Jamie Taylor. 2019. Industry-scale Knowledge Graphs: Lessons and Challenges: Five diverse technology companies show how it’s done. Queue 17, 2 (2019), 48–75

    2019

  23. [31]

    Jeff Z Pan, Simon Razniewski, Jan-Christoph Kalo, Sneha Singhania, Jiaoyan Chen, Stefan Dietze, Hajira Jabeen, Janna Omeliyanenko, Wen Zhang, Matteo Lissandrini, et al . 2023. Large language models and knowledge graphs: Opportunities and challenges. arXiv preprint arXiv:2308.0...

    2023 arXiv

  24. [32]

    Adil Rasheed, Omer San, and Trond Kvamsdal. 2020. Digital twin: Values, challenges and enablers from a modeling perspective. IEEE access 8 (2020), 21980–22012

    2020

  25. [33]

    Chaouki Saidi, Nadia Hamani, Mounir Benaissa, Benjamin Rolf, Tobias Reggelin, and Sebastian Lang. 2025. Modeling reconfigurable supply chains using knowledge graphs: towards Supply Chain 5.0. Production Engineering (2025), 1–24

    2025

  26. [34]

    Sithursan Sivasubramaniam, Cedric E Osei-Akoto, Yi Zhang, Kurt Stockinger, and Jonathan Fürst. 2024. Sm3-text-to-query: Synthetic multi-model medical text-to-query benchmark. Advances in Neural Information Processing Systems 37 (2024), 88627–88663

    2024

  27. [35]

    Enoch Oluwademilade Sodiya, Uchenna Joseph Umoga, Olukunle Oladipupo Amoo, and Akoh Atadoga. 2024. Ai-driven warehouse automation: a com- prehensive review of systems. GSC Advanced Research and Reviews, 18 (2), 272-282

    2024

  28. [36]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems 35 (2022), 24824–24837. https://papers.nips.cc/paper_files/pa...

    2022

  29. [37]

    Yuchen Xia, Daniel Dittler, Nasser Jazdi, Haonan Chen, and Michael Weyrich

  30. [38]

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. arXiv preprint arXiv:2210.03629 (2022). doi:10.48550/ arXiv.2210.03629

    2022 doi arXiv

  31. [39]

    Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al. 2023. A survey of large language models. arXiv preprint arXiv:2303.18223 1, 2 (2023)

    2023 arXiv

  32. [40]

    Zhiheng Zhao, Mengdi Zhang, Jian Chen, Ting Qu, and George Q Huang

  33. [41]

    Ray Y Zhong, Xun Xu, Eberhard Klotz, and Stephen T Newman. 2017. Intel- ligent manufacturing in the context of industry 4.0: a review. Engineering 3, 5 (2017), 616–630

    2017

  34. [42]

    EvergreenEdge

    Yunqi Zou, Yongli Wang, and Dongmei Liu. 2024. Q2Cypher: Converting Natural Language Questions to Cypher with Fine-Tuned Large Language Models. In 2024 5th International Conference on Artificial Intelligence and Computer Engineering (ICAICE). IEEE, 783–788. A PREPRINT A Append...

    2024

  35. [2018]

    International Journal of Production Research 56, 8 (2018), 2753–2768

    Design and application of Internet of things-based warehouse manage- ment system for smart logistics. International Journal of Production Research 56, 8 (2018), 2753–2768

    2018

  36. [2019]

    Digital twin-driven manufacturing cyber-physical system for parallel controlling of smart workshop.Journal of ambient intelligence and humanized computing 10 (2019), 1155–1166

    2019

  37. [2022]

    Computers & Industrial Engineering 171 (2022), 108454

    Digital twin-enabled dynamic spatial-temporal knowledge graph for production logistics resource allocation. Computers & Industrial Engineering 171 (2022), 108454

    2022

  38. [2024]

    In 2024 IEEE 29th International Conference on Emerging Technologies and Factory Automation (ETFA)

    LLM experiments with simulation: Large Language Model Multi- Agent System for Simulation Model Parametrization in Digital Twins. In 2024 IEEE 29th International Conference on Emerging Technologies and Factory Automation (ETFA). IEEE, 1–4

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.