Pith. sign in

REVIEW 3 major objections 5 minor 84 references

AI Simulation by Digital Twins: Systematic Survey, Reference Framework, and Mapping to a Standardized Architecture

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

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

Pith's one-line read A study of 22 systems derives a reference framework, DT4AI, that gives digital twin AI simulation a common structure and an ISO 23247 mapping.

desk verdict A methodologically careful systematic survey of a small niche; the DT4AI framework and ISO 23247 mapping are useful contributions, but the 22-study corpus is too narrow to support strong generalizability claims. read the letter →

arxiv 2506.06580 v2 pith:2MIP2HV5 submitted 2025-06-06 cs.AI cs.ETcs.SEcs.SYeess.SY

Xiaoran Liu , Istvan David This is my paper
classification cs.AIcs.ETcs.SEcs.SYeess.SY
keywords digitaltwinsAIsimulationreferenceframeworkISO23247systematicsurveyreinforcementlearningsim-to-realtransferarchitecturemapping
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

The paper claims that digital twin-enabled AI simulation—training AI agents on data generated by high-fidelity virtual replicas of physical systems—has enough structure to be captured in a common reference framework. From 22 primary studies, it derives DT4AI, a framework with three components (the AI agent, the Digital Twin, and the Physical Twin) and seven labeled interactions that recur across systems. It then maps those components and interactions onto the ISO 23247 reference architecture for digital twins, turning the conceptual framework into architectural guidance. The survey also finds that most sampled systems are autonomous digital twins, that deep reinforcement learning dominates the AI techniques, and that networks and robotics account for over three-quarters of applications. A shared vocabulary and blueprint of this kind would let researchers and engineers compare, design, and standardize such systems instead of reinventing their architecture each time.

What carries the argument

The key machinery is the DT4AI reference framework itself: three components—AI agent, Digital Twin, Physical Twin—and seven named interactions (A Query, B Simulated data, C Observe, D Real data, E Update, F Control, G Direct access). The framework is what carries the argument: it is the decomposition schema used to compare all primary studies and the structure that is then aligned with ISO 23247's functional entities. Its variation points (explicit/implicit query, batch/live training, passive/active observation, synchronous/asynchronous update, in-place/deploy-and-control control) are what let the framework express both reinforcement-learning-style live interaction and deep-learning-style batch data delivery.

What would settle it

If a future systematic replication ran a broader automated search using synonyms such as 'virtual training environment' and 'simulation-based training' alongside 'digital twin,' and found a substantial number of additional primary studies that cannot be expressed as the seven DT4AI interactions or that overturn the reported domain distribution, the framework's claim to cover the state of the art would be refuted.

Watch

Extended reading notes

Core claim

The central discovery is that the apparent variety in digital twin-based AI simulation can be decomposed into a small set of recurring architectural units, which the paper organizes as the DT4AI framework. Each system is seen as an AI agent, a Digital Twin, and a Physical Twin, connected by a Query, Simulated data, Observe, Real data, Update, Control, and Direct access link, with variation points such as batch versus live training, passive versus active observation, and synchronous versus asynchronous update. The paper shows that all 22 sampled studies instantiate this scheme, and that each DT4AI element has a counterpart in ISO 23247's entities and functional entities, with the simulator mapping to the Simulation FE, the model to the Digital Representation FE, and data collection and control to the corresponding data-collection and device-control FEs. The mapping also exposes gaps in the standard, including missing support for data storage, advanced analytics and AI-specific computation, verification and validation (VVUQ), lifecycle management, and fuller security requirements.

Load-bearing premise

The load-bearing premise is that the 22-study corpus is representative of the state of the art in digital twin-enabled AI simulation; only four primary studies came from the automated database search, with the rest added by manual search, expert knowledge, and snowballing, so a biased sample would skew the trends and the framework derived from them.

Editorial extensions

If this is right

  • Any digital twin system built for AI simulation can be described in terms of AI, Digital Twin, and Physical Twin components and the seven interactions, enabling direct comparison across systems and domains.
  • ISO 23247 can serve as a development target: systems can place their simulator in the Simulation FE, their model in the Digital Representation FE, and their data and control links in the corresponding data-collection and device-control FEs.
  • Because reinforcement learning dominates the sampled work, digital twin infrastructures should expect many small, low-latency query-response cycles in live training in addition to occasional large batch data deliveries.
  • The rarity of on-demand simulator updating points to richer maintenance and lifecycle patterns for simulators as an open engineering opportunity, particularly on-demand and synchronous update mechanisms.
  • The identified gaps in ISO 23247—data storage, advanced analytics, VVUQ, ontology, lifecycle management, and security—define a concrete extension agenda for the standard.

Reading between the lines

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

  • One implication not drawn in the paper is that the seven interactions and their variation points could serve as a classification schema for future benchmarks, correlating choices like batch versus live training with training-data quality or sim-to-real transfer success.
  • A testable extension would be to apply DT4AI to a new corpus of digital twin AI simulation systems outside networks and robotics; any system requiring an interaction beyond the seven named ones would indicate the framework needs revision.
  • The gap analysis implies that a standards extension for AI simulation would need to add an AI-specific functional entity, a data-storage entity, and a verification, validation, and uncertainty quantification entity to ISO 23247, an agenda the paper gestures at but does not draft.
  • If the observed 86.4% share of genuine autonomous digital twins holds outside the sample, AI simulation may be an early adopter of advanced twinning rather than a follower of manufacturing-style digital shadows.
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 / 5 minor

Summary. The paper reports a systematic literature survey of digital twin-enabled AI simulation, analyzing 22 primary studies. It identifies trends in application domains, DT architectural patterns, AI/ML techniques, simulator lifecycle models, and supporting technologies. Based on the corpus, the authors derive the DT4AI conceptual reference framework, which positions an AI component, a Digital Twin, and a Physical Twin, connected by interactions A–G (query, simulated data, observation, real data, update, control, access control). They map DT4AI onto the ISO 23247-2:2021 reference architecture, identify gaps in the standard, and discuss challenges and research opportunities. A replication package is provided.

Significance. If its claims hold, the paper offers a useful common vocabulary and architectural blueprint for a young interdisciplinary field. Its strengths include a documented protocol with two snowballing rounds, inter-rater agreement measures, a quality assessment, and a publicly available replication package. The DT4AI framework and its mapping to ISO 23247 are actionable for practitioners, and the critical reflection in Section 7 identifies concrete gaps (e.g., storage, VVUQ, lifecycle management, security). The main limitation is the small, potentially non-representative corpus, so the framework's generality is not fully established; this tempers but does not erase the contribution.

major comments (3)
  1. [§3.2, §4] Section 4 states that the authors 'sampled a statistically adequate corpus' and then uses sample-based generalization to derive DT4AI. This claim is not supported. Section 3.2.1 reports that the automated search found only 4 primary studies, with the remaining 18 added via manual search, expert knowledge, and snowballing; Section 3.2.7 acknowledges potential selection bias. The corpus is also heavily weighted toward networks and robotics (17 of 22, 77.3%, per Section 6.1.2). Please provide a formal justification for statistical adequacy (e.g., saturation analysis across strata or confidence intervals on reported frequencies), or explicitly reframe the framework as an evidence-based synthesis of the sampled studies rather than a statistically generalizable reference framework.
  2. [§4] The derivation of DT4AI is not transparent enough to assess its grounding. Section 4 says each primary study was decomposed into architectural units and recurring patterns identified, but no table or diagram maps each framework component or interaction (A–G) to the supporting primary studies. Without such traceability, a reader cannot distinguish corpus-grounded patterns from a priori schema. Please add a mapping table linking each component and variation point to the corresponding PS entries, or explicitly point to the relevant sheet of the replication package and summarize that mapping in the paper.
  3. [§7, Table 15] The mapping to ISO 23247 uses 'strongly related' and 'partially related' without defining the criteria for these classifications. Since this mapping is the basis for the paper's architectural guidelines, the reader needs to know whether the strength of relation follows systematically from the ISO definitions, from corpus evidence, or from expert judgment. Please specify the classification procedure and, ideally, report inter-rater agreement for the mapping as was done for study screening. If the mapping is intended as a proposal rather than a finding, state that explicitly.
minor comments (5)
  1. [§3.2.2] The manual search window is stated as 2019–2024, yet the final corpus includes PS17 (2018) and PS5 (2019); clarify that snowballing and other search phases can add studies outside this window.
  2. [§3.2.4] The claim that snowballing was applied 'until saturation' is overstated when only two rounds were conducted; the reported low inclusion rate is not direct evidence of saturation. Please rephrase.
  3. [Table 2] Variation points such as 'Big data' vs. 'Small data' and 'Low-context' vs. 'High-context' are qualitative and would benefit from brief operational definitions or examples.
  4. [§3, §5] The paper does not disclose that primary study PS3 and the taxonomy in [17] are authored by the second author; add a transparency note so readers can weigh potential self-reference effects.
  5. [§3.3] There is a typo in the sentence 'This is score is due to...' — it should read 'This score is due to...'.

Circularity Check

0 steps flagged · score 2.0 of 10

Minor self-citations present but no load-bearing circularity; the DT4AI framework is an inductive generalization from an acknowledged corpus and the ISO mapping is an external conceptual alignment.

full rationale

The core derivation, the DT4AI framework, is explicitly an inductive generalization: "we construct a conceptual reference framework from the sampled primary studies. We rely on a mixed sample- and case-based generalization [75]" (Section 4). The framework components A-G are abstractions of recurring interaction patterns reported in the 22 primary studies, not fitted parameters, and the paper does not claim to predict any quantity from a fitted input. The ISO 23247 mapping (Section 7) is a conceptual alignment against an external standard, accompanied by a critical reflection on the standard's gaps, so it does not reduce to the framework's own definitions. The only self-referential elements are primary study PS3, authored by the second author, and the use of David et al. [17], also co-authored by the second author, as part of the taxonomy "initial values" in Section 5.2. These are transparently disclosed and are not load-bearing: removing them would not change the A-G interaction patterns, which are grounded in the rest of the corpus, and the taxonomy merely labels DT variants rather than generating the framework. The weak automated-search yield (only 4 of 22 studies from automated search, with the remainder from manual search, expert knowledge, and snowballing) and the acknowledged potential selection bias are external-validity threats, not circularity. No equation is defined in terms of a target result, and no fitted value is renamed as a prediction.

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

The central claim rests on the representativeness of the 22-study corpus and on the suitability of existing taxonomies. No free parameters or invented physical entities are introduced; the DT4AI framework is a conceptual synthesis, not a postulated entity with falsifiable predictions.

assumptions (3)
  • domain assumption The 22 primary studies are representative of the state of the art in DT-enabled AI simulation.
    The entire analysis and framework derivation depend on the corpus reflecting the field; Section 3.2 documents the search, but the automated query returned only 4 studies, making representativeness an assumption.
  • domain assumption The taxonomies of Kritzinger et al. and David et al. are appropriate for classifying all sampled studies.
    Used in Section 5.2 to classify architectural styles; if these taxonomies do not fit, the reported distribution would change.
  • domain assumption The informal generalization from 22 studies to a reference framework is valid.
    Section 4 uses sample-and-case-based generalization; there is no formal proof that the framework covers all possible DT-AI simulation architectures.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AI Simulation by Digital Twins: Systematic Survey, Reference Framework, and Mapping to a Standardized Architecture." pith.science (2026). https://pith.science/paper/2MIP2HV5

@misc{pith2026250606580,
  author       = {Pith},
  title        = {Pith review of: AI Simulation by Digital Twins: Systematic Survey, Reference Framework, and Mapping to a Standardized Architecture},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2MIP2HV5}},
  note         = {Machine review of arXiv:2506.06580}
}
read the original abstract

Insufficient data volume and quality are particularly pressing challenges in the adoption of modern subsymbolic AI. To alleviate these challenges, AI simulation uses virtual training environments in which AI agents can be safely and efficiently developed with simulated, synthetic data. Digital twins open new avenues in AI simulation, as these high-fidelity virtual replicas of physical systems are equipped with state-of-the-art simulators and the ability to further interact with the physical system for additional data collection. In this article, we report on our systematic survey of digital twin-enabled AI simulation. By analyzing 22 primary studies, we identify technological trends and derive a reference framework to situate digital twins and AI components. Based on our findings, we derive a reference framework and provide architectural guidelines by mapping it onto the ISO 23247 reference architecture for digital twins. Finally, we identify challenges and research opportunities for prospective researchers.

Figures

Figures reproduced from arXiv: 2506.06580 by the authors.

Figure 1
Figure 1. reports the basic mappings of publication trends in our corpus. The number of publications shows an increasing trend, with a clear increase in publication output in 2023, constituting half of the corpus. The relatively low number of studies in 2024 is partly due to our study being conducted in Q2/2024 and possibly due to seasonal variations in area-specific publication trends (e.g., timing of conferences). We observ… view at source ↗
Figure 2
Figure 2. The DT4AI framework 4 The DT4AI Framework To integrate DTs, AI, and simulation, we construct a conceptual reference framework from the sam￾pled primary studies. We rely on a mixed sample￾and case-based generalization [75]. This approach is particularly useful when constructing middle-range theories that balance generality with practicality, such as engineering sciences. In Section 3, we sampled a statistically adequ… view at source ↗
Figure 3
Figure 3. RQ2: Architectural patterns form of reinforcement learning. Deep Reinforcement Learning (DRL, 13 of 22 – 59.1%) is a heavily favored choice, with more value-based methods (8 of 22 – 36.4%) than policy-based (5 of 22 – 22.7%) ones. A deeper look into the details reveals state-of-the-art AI algorithms. Among value-based deep reinforce￾ment learning, we typically find variants of Deep Q Networks [PS15]; in policy-based… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: RQ3: AI patterns (relevant components highlighted) [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: RQ4: Simulator lifecycle patterns (relevant components highlighted) [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

84 extracted references · 57 canonical work pages

  1. [17]

    Infonomics of Autonomous Digital Twins

    Istvan David and Dominik Bork. “Infonomics of Autonomous Digital Twins”. In: Advanced In- formation Systems Engineering. Vol. 14663. Lec- ture Notes in Computer Science. Springer, 2024, pp. 563–578. isbn: 978-3-031-61057-8. doi: 10. 1007/978-3-031-61057-8_33

    work page 2024

  2. [1]

    Data augmentation and deep learning methods in sound classification: A sys- tematic review

    Olusola O Abayomi-Alli, Robertas Damaˇ seviˇ cius, Atika Qazi, Mariam Adedoyin-Olowe, and San- jay Misra. “Data augmentation and deep learning methods in sound classification: A sys- tematic review”. In: Electronics 11.22 (2022), p. 3795

    2022

  3. [2]

    Digital twin security: a perspective on efforts from standard- ization bodies

    Cristina Alcaraz and Javier Lopez. “Digital twin security: a perspective on efforts from standard- ization bodies”. In: IEEE Security & Privacy 23.1 (2025), pp. 83–90

    2025

  4. [3]

    Amazon Mechanical Turk

    Amazon, Inc. Amazon Mechanical Turk. https: //www.mturk.com/. Accessed: 2024-06-24. 2012

    2024

  5. [4]

    AI-Assisted Human La- beling: Batching for Efficiency without Overre- liance

    Zahra Ashktorab et al. “AI-Assisted Human La- beling: Batching for Efficiency without Overre- liance”. In: Proc. ACM Hum.-Comput. Interact. 5.CSCW1 (2021). doi: 10.1145/3449163

    doi:10.1145/3449163 2021

  6. [5]

    Artifact Review and Badging Version 1.1

    Association for Computing Machinery. Artifact Review and Badging Version 1.1. https://www. acm.org/publications/policies/artifact- 26 X. Liu and I. David review - and - badging - current. Accessed: 2024-06-24. 2020

    2024

  7. [6]

    Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods and applications

    Shanaka Kristombu Baduge, Sadeep Thi- lakarathna, Jude Shalitha Perera, Mehrdad Arashpour, Pejman Sharafi, Bertrand Teo- dosio, Ankit Shringi, and Priyan Mendis. “Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods and applications”. In: Automation in Construction 141 (2022), p. 104440. issn: 0926-...

    arXiv 2022

  8. [7]

    Digital twin as risk-free experimentation aid for techno-socio-economic systems

    Souvik Barat, Vinay Kulkarni, Tony Clark, and Balbir Barn. “Digital twin as risk-free experimentation aid for techno-socio-economic systems”. In: Proceedings of the 25th Interna- tional Conference on Model Driven Engineering Languages and Systems . MODELS ’22. ACM, 2022, pp. 66–75. isbn: 9781450394666. doi: 10 . 1145 / 3550355 . 3552409. url: https : //do...

    arXiv 2022

Show all 84 references
  1. [8]

    A survey on data augmenta- tion for text classification

    Markus Bayer, Marc-Andr´ e Kaufhold, and Christian Reuter. “A survey on data augmenta- tion for text classification”. In: ACM Computing Surveys 55.7 (2022), pp. 1–39

    2022

  2. [9]

    Model-Driven Development of a Digital Twin for Injection Molding

    Pascal Bibow, Manuela Dalibor, Christian Hopmann, Ben Mainz, Bernhard Rumpe, David Schmalzing, Mauritius Schmitz, and Andreas Wortmann. “Model-Driven Development of a Digital Twin for Injection Molding”. In: Advanced Information Systems Engineering . Springer, 2020, pp. 85–100...

    2020

  3. [10]

    Model va- lidity and tolerance quantification for real-time adaptive approximation

    Raheleh Biglari and Joachim Denil. “Model va- lidity and tolerance quantification for real-time adaptive approximation”. In: Proceedings of the 25th International Conference on Model Driven Engineering Languages and Systems: Compan- ion Proceedings . MODELS ’22. ACM, 2022, pp....

    2022

  4. [11]

    Pytorch vs Tensorflow: A Head-to-Head Comparison

    Gaudenz Boesch. Pytorch vs Tensorflow: A Head-to-Head Comparison. https://viso.ai/ deep - learning / pytorch - vs - tensorflow/. Accessed: 2024-10-23. 2023

    2024

  5. [12]

    Digital twin–the simulation aspect

    Stefan Boschert and Roland Rosen. “Digital twin–the simulation aspect”. In: Mechatronic futures. Springer, 2016, pp. 59–74

    2016

  6. [13]

    Closing the sim-to- real loop: Adapting simulation randomization with real world experience

    Yevgen Chebotar, Ankur Handa, Viktor Makoviychuk, Miles Macklin, Jan Issac, Nathan Ratliff, and Dieter Fox. “Closing the sim-to- real loop: Adapting simulation randomization with real world experience”. In: 2019 Interna- tional Conference on Robotics and Automation (ICRA). IEE...

    2019

  7. [14]

    Generative adversarial net- works: An overview

    Antonia Creswell, Tom White, Vincent Du- moulin, Kai Arulkumaran, Biswa Sengupta, and Anil A Bharath. “Generative adversarial net- works: An overview”. In: IEEE signal processing magazine 35.1 (2018), pp. 53–65

    2018

  8. [15]

    Opinion- Guided Reinforcement Learning

    Kyanna Dagenais and Istvan David. Opinion- Guided Reinforcement Learning. Tech. rep. 2024. arXiv: 2405.17287 [cs.LG]

    2024 arXiv

  9. [16]

    A Cross-Domain Systematic Mapping Study on Software Engineering for Digital Twins

    Manuela Dalibor, Nico Jansen, Bernhard Rumpe, David Schmalzing, Louis Wacht- meister, Manuel Wimmer, and Andreas Wortmann. “A Cross-Domain Systematic Mapping Study on Software Engineering for Digital Twins”. In: J Syst Softw 193 (2022), p. 111361. issn: 0164-1212. doi: https :...

    2022

  10. [19]

    Inference of Simulation Models in Digital Twins by Reinforcement Learning

    Istvan David, Jessie Galasso, and Eugene Syri- ani. “Inference of Simulation Models in Digital Twins by Reinforcement Learning”. In: 2021 ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C) . 2021, pp. 221–224. doi: 10.1109/...

    2021

  11. [20]

    Automated Inference of Simulators in Digital Twins

    Istvan David and Eugene Syriani. “Automated Inference of Simulators in Digital Twins”. In: Handbook of Digital Twins . CRC Press, 2024. Chap. 8, pp. 122–148. isbn: 9781032546070. doi: 10.1201/9781003425724-11

    2024 doi

  12. [21]

    Increasing the Speed and Accuracy of Data Labeling Through an AI Assisted Interface

    Michael Desmond et al. “Increasing the Speed and Accuracy of Data Labeling Through an AI Assisted Interface”. In: Proc. of the 26th International Conference on Intelligent User Interfaces. ACM, 2021, pp. 392–401. isbn: 9781450380171

    2021

  13. [22]

    Sim2real transfer learning for 3D human pose estimation: AI Simulation by Digital Twins 27 motion to the rescue

    Carl Doersch and Andrew Zisserman. “Sim2real transfer learning for 3D human pose estimation: AI Simulation by Digital Twins 27 motion to the rescue”. In: Advances in Neural Information Processing Systems. Vol. 32. Curran Associates, Inc., 2019

    2019

  14. [23]

    Reference Archi- tectural Model Industrie 4.0 (RAMI4.0) – An Introduction

    Federal Ministry for Economic Affairs and Germany Climate Action. Reference Archi- tectural Model Industrie 4.0 (RAMI4.0) – An Introduction . https : / / www . plattform - i40 . de / IP / Redaktion / EN / Downloads / Publikation/rami40-an-introduction.html . Accessed: 2024-06-24. 2018

    2024

  15. [24]

    Awesome Digital Twins – In- teresting Papers

    EDT Community. Awesome Digital Twins – In- teresting Papers . https : / / github . com / edt - community / awesome - digital - twins ? tab = readme - ov - file # interesting - papers. Ac- cessed: 2024-06-24. 2024

    2024

  16. [25]

    A survey of data quality measurement and monitoring tools

    Lisa Ehrlinger and Wolfram W¨ oß. “A survey of data quality measurement and monitoring tools”. In: Frontiers in big data 5 (2022), p. 850611

    2022

  17. [26]

    What Is the Role of AI for Digital Twins?

    Frank Emmert-Streib. “What Is the Role of AI for Digital Twins?” In: AI 4.3 (2023), pp. 721–

    2023

  18. [27]

    Smart and collaborative industrial IoT: A federated learning and data space approach

    Bahar Farahani and Amin Karimi Monsefi. “Smart and collaborative industrial IoT: A federated learning and data space approach”. In: Digital Communications and Networks 9.2 (2023), pp. 436–447

    2023

  19. [28]

    Standardisation in digital twin architectures in manufacturing

    Enxhi Ferko, Alessio Bucaioni, Patrizio Pellic- cione, and Moris Behnam. “Standardisation in digital twin architectures in manufacturing”. In: 2023 IEEE 20th International Conference on Software Architecture (ICSA) . IEEE. 2023, pp. 70–81

    2023

  20. [29]

    Industry 4.0 Asset Ad- ministration Shell

    OPC Foundation. Industry 4.0 Asset Ad- ministration Shell . https : / / reference . opcfoundation . org / I4AAS / v100 / docs. Accessed: 2024-06-24. 2021

    2024

  21. [30]

    Big data preprocessing: methods and prospects

    Salvador Garc ´ ıa, Sergio Ram ´ ırez-Gallego, Juli´ an Luengo, Jos´ e Manuel Ben ´ ıtez, and Francisco Herrera. “Big data preprocessing: methods and prospects”. In: Big data analytics 1 (2016), pp. 1–22

    2016

  22. [31]

    Innovation Insight: AI Simulation

    Gartner Research. Innovation Insight: AI Simulation. https : / / www . gartner . com / en / documents / 4037399. Accessed: 2024-06-24. 2023

    2024

  23. [32]

    Ef- fectiveness and efficiency of search methods in systematic reviews of complex evidence: audit of primary sources

    Trisha Greenhalgh and Richard Peacock. “Ef- fectiveness and efficiency of search methods in systematic reviews of complex evidence: audit of primary sources”. In: BMJ 331.7524 (2005), pp. 1064–1065. issn: 0959-8138

    2005

  24. [33]

    15 challenges for AI: or what AI (currently) can’t do

    Thilo Hagendorff and Katharina Wezel. “15 challenges for AI: or what AI (currently) can’t do”. In: AI & SOCIETY 35.2 (2020), pp. 355–

    2020

  25. [34]

    Publish or Perish

    Anne-Wil Harzing. Publish or Perish . https:// harzing.com/resources/publish-or-perish . Accessed: 2024-06-24. 2016

    2024

  26. [35]

    How Simulation Helps Autonomous Driving: A Survey of Sim2real, Digital Twins, and Parallel Intelligence

    Xuemin Hu, Shen Li, Tingyu Huang, Bo Tang, Rouxing Huai, and Long Chen. “How Simulation Helps Autonomous Driving: A Survey of Sim2real, Digital Twins, and Parallel Intelligence”. In: IEEE Transactions on Intel- ligent Vehicles 9.1 (2024), pp. 593–612. doi: 10.1109/TIV.2023.3312777

    2024

  27. [36]

    Digital Twins of Manufacturing Systems as a Base for Machine Learning

    Florian Jaensch, Akos Csiszar, Christian Scheifele, and Alexander Verl. “Digital Twins of Manufacturing Systems as a Base for Machine Learning”. In: 25th International Conference on Mechatronics and Machine Vision in Practice . 2018, pp. 1–6

    2018

  28. [37]

    Sim2Real Predictivity: Does Evaluation in Simulation Predict Real-World Performance?

    Abhishek Kadian, Joanne Truong, Aaron Gokaslan, Alexander Clegg, Erik Wijmans, Stefan Lee, Manolis Savva, Sonia Chernova, and Dhruv Batra. “Sim2Real Predictivity: Does Evaluation in Simulation Predict Real-World Performance?” In: IEEE Robotics and Automa- tion Letters 5.4 (202...

    2020

  29. [38]

    Edge Computing-Based Digital Twin Frame- work Based on ISO 23247 for Enhancing Data Processing Capabilities

    Min-Su Kang, Dong-Hee Lee, Mahdi Sadeqi Bajestani, Duck Bong Kim, and Sang Do Noh. “Edge Computing-Based Digital Twin Frame- work Based on ISO 23247 for Enhancing Data Processing Capabilities”. In: Machines 13.1 (2024), p. 19

    2024

  30. [39]

    Meta-sim: Learning to generate synthetic datasets

    Amlan Kar, Aayush Prakash, Ming-Yu Liu, Eric Cameracci, Justin Yuan, Matt Rusiniak, David Acuna, Antonio Torralba, and Sanja Fi- dler. “Meta-sim: Learning to generate synthetic datasets”. In: Proceedings of the IEEE/CVF International Conference on Computer Vision . 2019, pp. 4551–4560

    2019

  31. [40]

    Guidelines for performing systematic literature reviews in software engineering

    Barbara Kitchenham and Stuart Charters. Guidelines for performing systematic literature reviews in software engineering . Tech. rep. EBSE-2007-01. Keele, UK, Keele University, 2007, pp. 1–65

    2007

  32. [41]

    Audio augmenta- tion for speech recognition

    Tom Ko, Vijayaditya Peddinti, Daniel Povey, and Sanjeev Khudanpur. “Audio augmenta- tion for speech recognition.” In: Interspeech. Vol. 2015. 2015, p. 3586

    2015

  33. [42]

    Dig- 28 X. Liu and I. David ital Twin in manufacturing: A categorical literature review and classification

    Werner Kritzinger, Matthias Karner, Georg Traar, Jan Henjes, and Wilfried Sihn. “Dig- 28 X. Liu and I. David ital Twin in manufacturing: A categorical literature review and classification”. In: IF AC-PapersOnLine 51.11 (2018). 16th IF AC Symposium on Information Control Proble...

    2018 doi

  34. [43]

    Constructing Neural Network Based Models for Simulating Dynamical Sys- tems

    Christian Legaard, Thomas Schranz, Gerald Schweiger, J´ an Drgoˇ na, Basak Falay, Cl´ audio Gomes, Alexandros Iosifidis, Mahdi Abkar, and Peter Larsen. “Constructing Neural Network Based Models for Simulating Dynamical Sys- tems”. In: ACM Comput. Surv. 55.11 (2023). issn: 0360...

    2023 doi

  35. [44]

    AI Simulation by Digital Twins: Systematic Survey of the State of the Art and a Reference Framework

    Xiaoran Liu and Istvan David. “AI Simulation by Digital Twins: Systematic Survey of the State of the Art and a Reference Framework”. In: ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Companion, MODELS-C . 1st Inter- national Conference on ...

    2024

  36. [45]

    Digital Twin Evolution for Sustain- able Smart Ecosystems

    Judith Michael, Istvan David, and Dominik Bork. “Digital Twin Evolution for Sustain- able Smart Ecosystems”. In: ACM/IEEE International Conference on Model Driven En- gineering Languages and Systems Companion, MODELS-C. ACM, 2024, pp. 1061–1065. doi: 10.1145/3652620.3688343

    2024

  37. [46]

    Digital Twins: A Survey on Enabling Technologies, Challenges, Trends and Future Prospects

    Stefan Mihai et al. “Digital Twins: A Survey on Enabling Technologies, Challenges, Trends and Future Prospects”. In: IEEE Communications Surveys & Tutorials 24.4 (2022), pp. 2255–2291. doi: 10.1109/COMST.2022.3208773

    2022

  38. [47]

    Towards an Ontological Framework for Validity Frames

    Rakshit Mittal, Raheleh Eslampanah, Lu- cas Lima, Hans Vangheluwe, and Dominique Blouin. “Towards an Ontological Framework for Validity Frames”. In: 2023 ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Compan- ion (MODELS-C) . 2023, pp. 801–...

    2023

  39. [48]

    Current trends in digital twin development, maintenance, and operation: an in- terview study

    Hossain M. Muctadir, David A. Manrique Negrin, Raghavendran Gunasekaran, Loek Cleophas, Mark van den Brand, and Boudewijn R. Haverkort. “Current trends in digital twin development, maintenance, and operation: an in- terview study”. In: Soft. Sys. Mod. (2024). issn: 1619-1374. ...

    2024 doi

  40. [49]

    Data augmentation: A comprehensive survey of mod- ern approaches

    Alhassan Mumuni and Fuseini Mumuni. “Data augmentation: A comprehensive survey of mod- ern approaches”. In: Array 16 (2022), p. 100258

    2022

  41. [50]

    Is Seeing Still Believ- ing? Leveraging Deepfake Technology for Livestock Farming

    Suresh Neethirajan. “Is Seeing Still Believ- ing? Leveraging Deepfake Technology for Livestock Farming”. In: Frontiers in Vet- erinary Science 8 (2021). issn: 2297-1769. doi: 10 . 3389 / fvets . 2021 . 740253. url: https : / / www . frontiersin . org / articles / 10.3389/fvets...

    2021

  42. [51]

    Semi-automatic semantic an- notation of PubMed queries: a study on quality, efficiency, satisfaction

    Aur´ elie N´ ev´ eol, Rezarta Islamaj Do˘ gan, and Zhiyong Lu. “Semi-automatic semantic an- notation of PubMed queries: a study on quality, efficiency, satisfaction”. In: Journal of biomedical informatics 44.2 (2011), pp. 310–318

    2011

  43. [52]

    ARIMA model building and the time series analysis approach to forecast- ing

    Paul Newbold. “ARIMA model building and the time series analysis approach to forecast- ing”. In: Journal of Forecasting 2.1 (1983), pp. 23–35. doi: https://doi.org/10.1002/ for.3980020104

    1983

  44. [53]

    Hu- man vs ChatGPT: Effect of Data Annotation in Interpretable Crisis-Related Microblog Classification

    Thi Huyen Nguyen and Koustav Rudra. “Hu- man vs ChatGPT: Effect of Data Annotation in Interpretable Crisis-Related Microblog Classification”. In: Proceedings of the ACM on Web Conference 2024 . WWW ’24. ACM, 2024, pp. 4534–4543. isbn: 9798400701719. doi: 10.1145/3589334.3648141

    2024

  45. [54]

    COOCK project Smart Port 2025 D3.1: ”To Twin Or Not To Twin”

    Randy Paredis, Hans Vangheluwe, and Pamela Adelino Ramos Albertins. COOCK project Smart Port 2025 D3.1: ”To Twin Or Not To Twin”. 2024. arXiv: 2401.12747

    2025 arXiv

  46. [55]

    A sur- vey on hardware accelerators: Taxonomy, trends, challenges, and perspectives

    Biagio Peccerillo, Mirco Mannino, Andrea Mondelli, and Sandro Bartolini. “A sur- vey on hardware accelerators: Taxonomy, trends, challenges, and perspectives”. In: Journal of Systems Architecture 129 (2022), p. 102561. issn: 1383-7621. doi: https : //doi.org/10.1016/j.sysarc.2...

    2022

  47. [56]

    A generic model for sustainability with process- and product-specific instances

    Birgit Penzenstadler and Henning Femmer. “A generic model for sustainability with process- and product-specific instances”. In: Proceedings of the 2013 Workshop on Green in/by Software Engineering. GIBSE ’13. Fukuoka, Japan: ACM, 2013, pp. 3–8. isbn: 9781450318662. doi: 10 . 1...

    2013

  48. [57]

    Systematic mapping studies in software engineering

    Kai Petersen, Robert Feldt, Shahid Mujtaba, and Michael Mattsson. “Systematic mapping studies in software engineering”. In: Proceedings of the 12th International Conference on Evalu- ation and Assessment in Software Engineering . AI Simulation by Digital Twins 29 EASE’08. Ital...

    2008

  49. [58]

    Guidelines for conducting systematic mapping studies in software engineering: An up- date

    Kai Petersen, Sairam Vakkalanka, and Ludwik Kuzniarz. “Guidelines for conducting systematic mapping studies in software engineering: An up- date”. In: Inf Softw Technol 64 (2015), pp. 1–

    2015

  50. [59]

    Digital Twin and Big Data Towards Smart Manufacturing and Indus- try 4.0: 360 Degree Comparison

    Qinglin Qi and Fei Tao. “Digital Twin and Big Data Towards Smart Manufacturing and Indus- try 4.0: 360 Degree Comparison”. In: IEEE Ac- cess 6 (2018), pp. 3585–3593. doi: 10 . 1109 / ACCESS.2018.2793265

    2018

  51. [60]

    Simulation

    Sheldon M Ross. Simulation. en. 6th ed. San Diego, CA: Academic Press, 2022

    2022

  52. [61]

    doi: https : / / doi

    issn: 0950-5849. doi: https : / / doi . org / 10.1016/j.infsof.2015.03.007

    2015 doi

  53. [62]

    Use case scenarios for digital twin implementation based on ISO 23247

    Guodong Shao. Use case scenarios for digital twin implementation based on ISO 23247 . Tech. rep. 2021. doi: 10.6028/nist.ams.400-2

    2021 doi

  54. [63]

    Digital Twins for Advanced Manu- facturing: The Standardized Approach

    Guodong Shao, Deogratias Kibira, and Simon Frechette. “Digital Twins for Advanced Manu- facturing: The Standardized Approach”. In:Dig- ital Twins, Simulation, and the Metaverse: Driv- ing Efficiency and Effectiveness in the Physical World through Simulation in the Virtual Worl...

    2024

  55. [64]

    Bridging the Si- los of Digitalization and Sustainability by Twin Transition: A Multivocal Literature Review

    Baran Shajari and Istvan David. Bridging the Si- los of Digitalization and Sustainability by Twin Transition: A Multivocal Literature Review

  56. [65]

    Connectionist AI, symbolic AI, and the brain

    P. Smolensky. “Connectionist AI, symbolic AI, and the brain”. In: Artificial Intelligence Review 1.2 (1987), pp. 95–109. issn: 1573-7462. doi: 10. 1007/BF00130011. url: https://doi.org/10. 1007/BF00130011

    1987

  57. [66]

    Rein- forcement learning: An introduction

    Richard S Sutton and Andrew G Barto. Rein- forcement learning: An introduction. MIT press, 2018

    2018

  58. [67]

    Assessing the Ability of ChatGPT to Screen Articles for Systematic Reviews

    Eugene Syriani, Istvan David, and Gauransh Kumar. Assessing the Ability of ChatGPT to Screen Articles for Systematic Reviews . 2023. arXiv: 2307.06464

    2023 arXiv

  59. [68]

    A survey on image data augmentation for deep learning

    Connor Shorten and Taghi M Khoshgoftaar. “A survey on image data augmentation for deep learning”. In: Journal of big data 6.1 (2019), pp. 1–48

    2019

  60. [69]

    Sim-to-real: Learning agile locomotion for quadruped robots

    Jie Tan, Tingnan Zhang, Erwin Coumans, Atil Iscen, Yunfei Bai, Danijar Hafner, Steven Bohez, and Vincent Vanhoucke. “Sim-to-real: Learning agile locomotion for quadruped robots”. In: arXiv preprint arXiv:1804.10332 (2018)

    2018 arXiv

  61. [70]

    Development of Digital Twin for Load Center on the Example of Distribution Network of an Ur- ban District

    Nikita Tomin, Victor Kurbatsky, Vadim Borisov, and Sergey Musalev. “Development of Digital Twin for Load Center on the Example of Distribution Network of an Ur- ban District”. In: E3S Web of Conferences 209 (2020), p. 02029. issn: 2267-1242. doi: 10.1051/e3sconf/202020902029

    2020

  62. [71]

    Challenges in Implementing Digital Twins – A Survey

    Jakob Trauer, Michael Mutschler, Markus M¨ ortl, and Markus Zimmermann. “Challenges in Implementing Digital Twins – A Survey”. In: vol. 42nd Computers and Information in Eng Conf. Intl Design Engineering Technical Conf and Computers and Information in Engineering Conf. 2022

    2022

  63. [72]

    Screening articles for systematic reviews with ChatGPT

    Eugene Syriani, Istvan David, and Gauransh Kumar. “Screening articles for systematic reviews with ChatGPT”. In: Journal of Com- puter Languages 80 (2024), p. 101287. issn: 2590-1184. doi: https://doi.org/10.1016/j. cola.2024.101287

    2024

  64. [73]

    An introduction to multi- paradigm modelling and simulation

    Hans Vangheluwe, Juan de Lara, and Pieter J Mosterman. “An introduction to multi- paradigm modelling and simulation”. In: Proceedings of the AIS2002 Conference (AI, Simulation and Planning in High Autonomy Systems). 2002, pp. 9–20

    2002

  65. [74]

    Digital Twin Development and Operation of a Flexible Man- ufacturing Cell using ISO 23247

    Bernhard Wallner, Benedikt Zw¨ olfer, Thomas Trautner, and Friedrich Bleicher. “Digital Twin Development and Operation of a Flexible Man- ufacturing Cell using ISO 23247”. In: Procedia CIRP 120 (2023), pp. 1149–1154

    2023

  66. [75]

    Six strate- gies for generalizing software engineering the- ories

    Roel J. Wieringa and Maya Daneva. “Six strate- gies for generalizing software engineering the- ories”. In: Sci. Comput. Program. 101 (2015), pp. 136–152. doi: 10.1016/j.scico.2014.11. 013

    2015 doi

  67. [76]

    Validity Frame–enabled model-based en- gineering processes

    Bert Van Acker, Paul De Meulenaere, Hans Vangheluwe, and Joachim Denil. “Validity Frame–enabled model-based en- gineering processes”. In: SIMULATION 100.2 (2024), pp. 185–226. doi: 10 . 1177 / 00375497231205035. url: https://doi.org/ 10.1177/00375497231205035

    2024 doi

  68. [77]

    Guidelines for the search strategy to update systematic liter- ature reviews in software engineering

    Claes Wohlin, Emilia Mendes, Katia Romero Fe- lizardo, and Marcos Kalinowski. “Guidelines for the search strategy to update systematic liter- ature reviews in software engineering”. In: In- 30 X. Liu and I. David formation and Software Technology 127 (2020), p. 106366

    2020

  69. [78]

    An Adapted Model of Cognitive Digital Twins for Building Lifecycle Management

    Ibrahim Yitmen, Sepehr Alizadehsalehi, ˙Ilknur Akıner, and Muhammed Ernur Akıner. “An Adapted Model of Cognitive Digital Twins for Building Lifecycle Management”. In: Applied Sciences 11.9 (2021). issn: 2076-3417. doi: 10.3390/app11094276

    2021 doi

  70. [79]

    Theory of modeling and simulation: discrete event & iterative system computational foundations

    Bernard P Zeigler, Alexandre Muzy, and Ernesto Kofman. Theory of modeling and simulation: discrete event & iterative system computational foundations . Academic press, 2018

    2018

  71. [80]

    Guidelines for snowballing in systematic literature studies and a replication in software engineering

    Claes Wohlin. “Guidelines for snowballing in systematic literature studies and a replication in software engineering”. In: Proceedings of the 18th International Conference on Evaluation and Assessment in Software Engineering . EASE ’14. ACM, 2014. isbn: 9781450324762. doi: 10....

    2014

  72. [81]

    Machine learn- ing on big data: Opportunities and chal- lenges

    Lina Zhou, Shimei Pan, Jianwu Wang, and Athanasios V. Vasilakos. “Machine learn- ing on big data: Opportunities and chal- lenges”. In: Neurocomputing 237 (2017), pp. 350–361. issn: 0925-2312. doi: https : //doi.org/10.1016/j.neucom.2017.01.026 . url: https : / / www . scienced...

    2017 doi

  73. [84]

    Sim-to-Real Transfer in Deep Re- inforcement Learning for Robotics: a Survey

    Wenshuai Zhao, Jorge Pe˜ na Queralta, and Tomi Westerlund. “Sim-to-Real Transfer in Deep Re- inforcement Learning for Robotics: a Survey”. In: 2020 IEEE Symposium Series on Compu- tational Intelligence (SSCI) . 2020, pp. 737–744. doi: 10.1109/SSCI47803.2020.9308468

    2020

  74. [365]

    issn: 1435-5655. doi: 10 . 1007 / s00146 - 019-00886-y

  75. [728]

    doi: 10.3390/ai4030038

    issn: 2673-2688. doi: 10.3390/ai4030038. url: https://www.mdpi.com/2673-2688/4/3/ 38

    doi

  76. [2025]

    04267 [physics.soc-ph]

    arXiv: 2506 . 04267 [physics.soc-ph]. url: https://arxiv.org/abs/2506.04267

    arXiv

Pith tools

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