Textual-Based vs. Thinging Machines Conceptual Modeling

Sabah Al-Fedaghi

REVIEW 4 major objections 4 minor 15 references

Textual-Based vs. Thinging Machines Conceptual Modeling

T0 review · 4 major / 4 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 This paper claims that thinging-machine diagrams built from five generic actions plus time capture domains more faithfully than text or UML class diagrams.

desk verdict An honest but unevidenced pitch for the author's own thinging-machine notation; read it for the examples, not for the claimed superiority over UML or text. read the letter →

arxiv 2506.02646 v1 pith:2TKJ2COR submitted 2025-06-03 cs.SE

Sabah Al-Fedaghi This is my paper · ORCID

classification cs.SE

keywords conceptualmodelingthingingmachinethimactext-baseddiagrammaticstaticanddynamicmodelsUMLclassdiagramsfivegenericactions

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

Domain modeling turns a natural-language description into a conceptual model, and the field splits into textual, diagrammatic, and mixed approaches. This paper argues that the thinging machine (TM) is the more appropriate vehicle for that translation: a TM model represents every entity and event as a thimac, a thing/machine whose behavior is limited to five generic actions (create, process, release, transfer, receive) together with a time feature that turns static structures into dynamic event chronologies. To support the argument, it develops three side-by-side examples (store sales, a volunteer pickup-and-delivery service, and an informal logic milk argument) and shows how each textual description can be re-expressed as a static TM diagram, a dynamic event model, and a chronology of events, with equivalent textual re-descriptions. The payoff if the claim holds is a conceptual modeling method with a narrow, unambiguous semantics that lets modelers filter and revise a text's meaning in a controlled, repeatable way.

What carries the argument

The central object is the thimac (thing/machine): a unit that is at once a thing and a machine, whose only capacities are the five actions create, process, release, transfer, and receive. Arrive and accept are collapsed into receive, and diagrams include storage and triggering. The machinery's work is twofold: at the static level, thimacs form regions that fix the structure and boundaries of a domain; at the dynamic level, the same thimacs are stamped with time to form events, and the chronology of those events is the model's account of behavior. The paper also uses a simplification rule, erasing release, transfer, and receive whenever flow direction is clear, to reduce diagram complexity, plus a multilevel reduction that lets create and process stand for whole sub-diagrams.

What would settle it

Take the H2S sentence 'A scheduled pickup may occur anytime between 8:00 and 14:00' and require a TM model to preserve both the time window and the 'may' possibility using only the five actions plus time. If the faithful model needs any mechanism beyond create, process, release, transfer, receive, or if erasing release, transfer, and receive changes whether the constraint can be expressed, the paper's simplification and generality claims fail.

Watch

Extended reading notes

Core claim

The central claim is that TM modeling is a more appropriate conceptual modeling methodology than the alternatives examined: purely textual descriptions, diagrammatic notations such as UML class diagrams, and mixed text-plus-diagram approaches. A TM model is defined over thimacs, which are simultaneously things and machines; each thimac's behavior consists of five generic actions (create, process, release, transfer, receive), and adding time lifts the static arrangement of thimacs into dynamic events whose chronology can itself be written as text. The paper shows, through the three worked domains, that the same source text can be mapped to a static TM region, a dynamic event carving, and a chronological event sequence, so that static, dynamic, and chronological views are systematically realigned versions of one underlying model. It further claims that the limited action vocabulary restricts the semantics and thereby limits ambiguity, and that the diagrams can be simplified by deleting release, transfer, and receive (or by treating diagrammatic notation as create and process) without losing the intended meaning.

Load-bearing premise

The load-bearing premise is that every domain description can be faithfully represented by the five actions create, process, release, transfer, receive plus a time dimension, and that simplifying a diagram by deleting release, transfer, and receive preserves its meaning; the paper itself leaves possibility words such as may be and could be unsolved.

Editorial extensions

If this is right

A modeler can translate a natural-language domain description into a static TM diagram, then derive a dynamic event model and a textual chronology from the same underlying thimacs, giving text and diagram roles that complement rather than compete.
If the simplification rule preserves meaning, large TM diagrams can be aggressively pruned by erasing release, transfer, and receive, which would make the notation practical for domains larger than a sheet of paper.
The five-action semantics gives each modeled statement a restricted interpretive space, so two modelers should converge on the same TM structure more often than with free-form UML class diagrams.
A TM diagram of a text gives modelers a filter-and-revise loop: information can be added or removed from the diagram in repeated passes, supporting iterative refinement of the domain description before design.

Reading between the lines

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

A testable extension the paper leaves implicit: take the H2S sentence 'A scheduled pickup may occur anytime between 8:00 and 14:00' and encode it purely with create, process, release, transfer, receive plus time; if faithful encoding requires an extra construct such as a clock or a modality, the five-action vocabulary is not by itself sufficient for the claimed generality.
If the TM simplification rule is sound, the same reduction could be applied to textual chronologies to yield a compact, machine-checkable event language, connecting TM modeling directly to model-to-text transformation and automated requirements analysis.
The paper's concluding admission that possibility words such as may be and could be remain unsolved suggests that the framework's natural next step is an explicit modality or possibility container; until then, informal arguments whose force depends on modal claims are only partially representable.

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.

Desk Editor's Note

read the letter

Colleague,

Short version: this is Al-Fedaghi applying his thinging machine (TM) framework to three new texts. The framework itself comes from his earlier papers; what is new is the application and the claim that TM is 'more appropriate' than UML or textual modeling. That claim is not supported, and the paper itself more or less admits it.

What the paper does well: the diagrams are careful, and attaching a chronology of events to a static TM model is a coherent extension of the author's program. The milk-argument example is a genuine attempt to apply TM to informal logic, which I have not seen in his earlier work. Credit also goes to the author's candor: Section IV says 'We cannot substantiate this claim except by putting the approaches side-by-side to contrast them,' and the conclusion acknowledges that 'may be, could be' remains unsolved. That is more honest than most papers with an unsupported central claim.

The soft spots are not subtle. The central claim is never evaluated against any explicit criterion — no completeness, no ambiguity, no modifiability, no user study. The examples are all constructed by the framework's creator, and the mapping from English to TM is unconstrained. Fig. 14 is the clearest sign: the model 'can be modified' to add the 8:00–14:00 pickup window, and other portions 'may not be necessary details.' The author decides what counts as relevant, which turns the side-by-side comparison into a demonstration, not a test. The modal and temporal gap is real, and it is exactly the kind of content that separates conceptual modeling notations.

The citation pattern is worth noting: the foundational axioms rest on refs [10] and [11], both self-citations. That is not a flaw by itself, but here the load-bearing machinery is taken from those papers and never independently evaluated in this work.

Who should read this? Someone tracking the TM research program, or someone looking for a worked example of how one might map natural-language constraints onto a fixed action ontology. As a standalone argument for methodological superiority, it fails. That said, the paper is coherent on its own terms and unusually honest about what it cannot do, so I would not call it unserious.

Recommendation: do not send this to peer review in its current form. It needs an evaluation criterion, a less permissive modeling procedure, and a discussion of how it handles modal and temporal constraints before the comparative claim can be taken up.

Referee Report

4 major / 4 minor

Summary. The paper compares the thinging machine (TM) modeling approach with textual and mixed textual/diagrammatic conceptual modeling. It presents three worked examples (a sales process, the Helping Hand Store domain, and an informal milk-consumption argument) and develops static TM models, dynamic models with event chronologies, and textual specifications of event sequences. The Abstract and Conclusion assert that TM modeling is a more appropriate methodology than the other diagrammatic schemes examined, while the body of the paper describes TM's five generic actions (create, process, release, transfer, receive) and a two-level static/dynamic ontology. The paper also discusses simplifying TM diagrams by deleting release, transfer, and receive when flow direction is indicated by arrows.

Significance. If the central comparative claim were established, the paper would offer a substantive contribution to conceptual modeling by proposing a unified static/dynamic representation with a small action set. The paper has clear strengths: it provides fully worked examples with detailed diagrams, it candidly states in Section IV that it cannot substantiate the superiority claim except by side-by-side contrast, and it explicitly acknowledges open issues such as representing 'may be' and 'could be.' However, as written, the paper does not provide systematic evidence for its central claim, and several load-bearing assumptions about expressiveness and semantic preservation are introduced without adequate justification. The contribution is therefore better characterized as an illustrative exposition of TM modeling than as a demonstrated comparative evaluation.

major comments (4)

[Section IV] The central claim that TM modeling is "a more appropriate methodology" is not supported by any systematic evaluation. The paper itself states, "We cannot substantiate this claim except by putting the approaches side-by-side to contrast them." A side-by-side display is not evidence of superiority unless the models are assessed against explicit criteria such as completeness, ambiguity, modifiability, or developer effectiveness. No such criteria or measurements are provided. The Abstract and Conclusion present the comparative superiority claim as established, which is inconsistent with the disclaimer in the body.
[Fig. 14 and Conclusion] The treatment of modal and temporal constraints undermines the generality claim. The Fig. 14 caption says an analysis "can modify the TM model by incorporating missing parts (e.g., picking up between 8:00 and 14:00)" and that other portions "may not be necessary details." This permits the modeler to drop precisely the content that would test expressiveness, such as the 8:00-14:00 time window and the modality in "may occur." The Conclusion concedes that representing "may be, could be" remains unsolved. Consequently, either TM cannot represent such constraints, in which case it is not general, or it can represent them but the paper does not show how, in which case the examples are incomplete evidence. Either way, the superiority claim is not established.
[Section II.A] The paper's foundational axioms are taken from the author's prior publications, and the examples are constructed by the same author. This creates a circularity concern: the method is evaluated by its creator on self-selected examples, with no independent coders, inter-rater reliability, or external validation. For example, the claim that "the world is divided into thimacs" and that the five actions are exhaustive is presented as an ontological commitment rather than as a hypothesis open to empirical or comparative testing. This circularity is not by itself fatal, but it means the reader cannot distinguish framework-internal consistency from demonstrated practical superiority.
[Section III.A] The simplification rule "If the arrows sufficiently indicate the flow's direction, Fig. 4 can be simplified by deleting release, transfer, and receive" is load-bearing but unproven. Deleting three of the five defined actions is a substantive semantic reduction: release, transfer, and receive carry distinctions about readiness, boundary crossing, and acceptance that are not obviously recoverable from arrows alone. The paper gives no formal argument or counterexample analysis showing that the simplification preserves model meaning in all cases, including loops, triggers, and negative events. Since the later dynamic models rely on this simplification, the soundness of the examples depends on this unstated assumption.

minor comments (4)

[Throughout] The text contains typographical errors, including "Casher" in Fig. 4, "Taxis" in Figs. 5 and 6, and "sorties" in Section I where "stories" appears intended.
[Reference [12]] The Larman reference lists page numbers as "pp. page-page," which is incomplete and should be corrected.
[Figures] Several figures (e.g., Figs. 4, 5, 10, 17) are extremely dense and appear to be reproduced at low resolution, making their numbered flow annotations difficult to follow. The red highlighting in Fig. 14 will be invisible in grayscale print.
[Section V] The milk example introduces a "negative event" NE19 without a definition or a reference to the prior work where it is established; readers unfamiliar with [10] and [11] will not know how a negative event is represented in the TM action ontology.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the TM comparison is under-supported evidence, not a self-referential derivation.

full rationale

The paper makes no quantitative prediction and derives no result from equations; its central claim is an evaluative comparison. The TM framework is restated as definitions in Section II ('For the sake of a self-contained paper, the following section includes a brief review of the TM model'), so the five-action ontology is an explicit starting point rather than a conclusion smuggled in from the examples. The three worked examples are drawn from external sources (Larman's sale use case, Chen et al.'s H2S text/class diagram, Reed et al.'s milk argument), and the paper constructs TM diagrams for them. The Fig. 14 caption does show that the modeler may add omitted parts or dismiss others ('An analysis can modify the TM model by incorporating missing parts (e.g., picking up between 8:00 and 14:00). Other portions may not be necessary details'), and the Conclusion concedes that 'representing may be, could be' remains unsolved. These admissions weaken the claim that TM is 'more appropriate' than UML or textual modeling, but they do not make the claim equivalent to its inputs by construction. The side-by-side comparison itself is just weak evidence, not a circular reduction; no fitted parameter is relabeled as a prediction, and no load-bearing premise is justified solely by an author self-citation. The self-citations to [10,11] supply background definitions, but the framework is restated and the comparison is carried out in this paper. Thus the correct circularity finding is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 2 invented entities

The paper introduces no free numeric parameters. Its load-bearing assumptions are the adequacy of the five-action ontology and the sufficiency of visual side-by-side comparison. All central constructs (thimac, five actions) originate in the author's prior self-cited work, giving the ledger a high degree of self-reference.

assumptions (4)

ad hoc to paper The world is divided into thimacs; all entities, properties, and relationships are thimacs or subthimacs.
This is the foundational ontology of TM, introduced and defended only in the author's prior work (refs [10], [11]); the paper provides no independent justification.
domain assumption Any textual domain description can be mapped to a TM diagram using the five generic actions (create, process, release, transfer, receive).
The paper demonstrates this only on three hand-picked examples and admits unresolved cases such as modal verbs (Conclusion).
ad hoc to paper Deleting release, transfer, and receive from TM diagrams when arrows indicate flow direction does not alter model semantics.
Stated without proof in Section III.A ('If the arrows sufficiently indicate the flow's direction, Fig. 4 can be simplified').
domain assumption Side-by-side visual comparison of text, UML, and TM is sufficient to judge methodological appropriateness.
The paper explicitly says it 'cannot substantiate this claim except by putting the approaches side-by-side' (Section IV), implying a subjective, non-quantitative criterion.

invented entities (2)

Thimac (thing/machine)
purpose: Central construct unifying things and machines; basis for all TM diagrams.
No falsifiable prediction or independent measurement is attached; it is a conceptual primitive from the author's prior papers.
Time feature forming dynamic events over static thimacs
purpose: Allows TM to represent event chronologies.
Conceptual device; no independent operationalization given.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Textual-Based vs. Thinging Machines Conceptual Modeling." pith.science (2026). https://pith.science/paper/2TKJ2COR

@misc{pith2026250602646,
  author       = {Pith},
  title        = {Pith review of: Textual-Based vs. Thinging Machines Conceptual Modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2TKJ2COR}},
  note         = {Machine review of arXiv:2506.02646}
}

read the original abstract

Software engineers typically interpret the domain description in natural language and translate it into a conceptual model. Three approaches are used in this domain modeling: textual languages, diagrammatic languages, and a mixed based of text and diagrams. According to some researchers, relying on a diagrammatic notation levies certain burdens for designing large models because visual languages are intended to depict everything diagrammatically during a development process but fail to do so for a lack of developer efficiency. It is claimed that textual formats enable easier manipulation in editors and tools and facilitate the integration of ontologies in software systems. In this paper, we explore the problem of the relationship between textual format and diagramming in conceptual modeling. The main focus is modeling based on the so-called thinging machine (TM). Several examples are developed in detail to contrast side-by-side targeted domains represented in textual description and TM modeling. A TM model is defined as a thimac (thing/machine) with a time feature that forms dynamic events over static thimacs utilizing five generic actions: create, process, release, transfer, and receive. This provides a conceptual foundation that can be simplified further by eliminating the actions of release, transfer, and receive. A multilevel reduction in the TM diagram s complexity can also be achieved by assuming diagrammatic notations represent the actions of creation and processing. We envision that special tools will help improve developer efficiency. The study s results of contrasting textual and mix-based descriptions vs. TM modeling justify our claim that TM modeling is a more appropriate methodology than other diagrammatic schemes (e.g., UML classes) examined in this paper.

Figures

Figures reproduced from arXiv: 2506.02646 by the authors.

**Figure 1.** Thimac. Create Receive Release Transfer Process Accept Arrive Output Input Existence (actuality) Subsistence (Potentiality) Events (regions + time) Regions (static Thimacs) [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 14 canonical work pages

[10]

Preconceptual modeling in software engineering: metaphysics of diagrammatic representations,

S. Al-Fedaghi, “Preconceptual modeling in software engineering: metaphysics of diagrammatic representations,” HAL archives,

work page
[11]

Exploring Conceptual Modeling Metaphysics: Existence Containers, Leibniz's Monads and Avicenna's Essence

S. Al-Fedaghi, “Exploring conceptual modeling metaphysics: existence containers, Leibniz’s monads and Avicenna’s essence,” arXiv preprint, arXiv:2405.01549, February 20, 2024

work page Pith review arXiv 2024
[1]

Conceptual modeling supported by text analysis,

B. Otsuka, “Conceptual modeling supported by text analysis,” Procedia Computer Science, vol. 126, pp. 1387-1394, 2018

work page 2018
[2]

Text-based modeling,

H. Gronniger, H. Krahn, B. Rumpe, M. Schindler, and S. Volkel, “Text-based modeling,” 4th International Workshop on Software Language Engineering (ateM 2007), Nashville, TN, USA, October 2007. https://www.researchgate.net/publication/266024095_Textbased _Modeling

work page arXiv 2007
[3]

Automated domain modeling with large language models: a comparative study,

K. Chen, Y. Yang, B. Chen, J. A. Hernandez Lopez, G. Mussbacher, and D. Varro, “Automated domain modeling with large language models: a comparative study,” ACM/IEEE 26th International Conference on Model Driven Engineering Languages and Systems (MODELS), Västerås, Sweden. October 1-6, 2023

work page 2023
[4]

Framework for conceptual modeling on natural language texts, 13-24,

M. Bogatyrev and K. Samodurov, “Framework for conceptual modeling on natural language texts, 13-24,” Third Workshop on Concept Discovery in Unstructured Data, Moscow, Russia, July 18, 2016. https://ceur-ws.org/Vol-1625/paper2.pdf

work page 2016
[5]

Leveraging LLMs in text-based ontology- driven conceptual modeling,

M. L. Coutinho, “Leveraging LLMs in text-based ontology- driven conceptual modeling,” July 15, 2024 - September 19, 2024, Enschede, Netherlands. https://www.utwente.nl/en/eemcs/fois2024/resources/papers/cout inho-leveraging-llms-in-text-based-ontology-driven-conceptual- modeling.pdf

work page 2024
[6]

Text-to-model transformation: natural language-based model generation framework

A. Akundi, J. Ontiveros, and S. Luna, “Text-to-model transformation: natural language-based model generation framework.” Systems, vol. 12, no. 369. https://doi.org/10.3390/systems12090369

doi:10.3390/systems12090369

Show all 15 references

[7]

The structure of the artistic text,

J. Lotman, “The structure of the artistic text,” Translated by Ronald Vroon, University of Michigan, Ann Arbor, 1977. https://monoskop.org/images/3/3e/Lotman_Jurij_The_Structure_ of_the_Artistic_Text_1977.pdf https://monoskop.org/images/3/3e/Lotman_Jurij_The_Structure_ of_the_...

1977
[8]

Media boundaries and conceptual modelling between texts and maps,

Ø. Eide, “Media boundaries and conceptual modelling between texts and maps,” Pre-print manuscript, April 2015. https://www.oeide.no/research/eideBetween.pdf

2015
[9]

Textual knowledge representation through the semantic-based graph structure in clustering applications,

J. Wu, Y. Dang, D. Pan, Z. Xuan, and Q. Liu, “Textual knowledge representation through the semantic-based graph structure in clustering applications,” 43rd Hawaii International Conference on System Sciences, January 5-8, 2010, HI, USA

2010
[12]

Larman, Applying UML and Patterns: An Introduction to Object-Oriented Analysis and Design and Iterative Development, 2nd ed., Prentice Hall, 2001, pp

C. Larman, Applying UML and Patterns: An Introduction to Object-Oriented Analysis and Design and Iterative Development, 2nd ed., Prentice Hall, 2001, pp. page-page. https://personal.utdallas.edu/~chung/SP/applying-uml-and- patterns.pd

2001
[13]

Model-to-text transformation: translating multiple conceptual models to user stories,

M. V. Plaetsen, “Model-to-text transformation: translating multiple conceptual models to user stories,” Master’s Dissertation, University of Ghent, Business Engineering, 2019- 2020. https://libstore.ugent.be/fulltxt/RUG01/002/837/851/RUG01- 002837851_2020_0001_AC.pdf

2019
[14]

Argument diagramming in logic, law and artificial intelligence,

C. Reed, D. Walton, and F. Macagno, “Argument diagramming in logic, law and artificial intelligence,” The Knowledge Engineering Review, vol. 22. No. 1, pp. 1-22, 2007

2007
[2024]

https://hal.archives-ouvertes.fr/hal-04576184

Pith tools

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

[1] [10]

Preconceptual modeling in software engineering: metaphysics of diagrammatic representations,

S. Al-Fedaghi, “Preconceptual modeling in software engineering: metaphysics of diagrammatic representations,” HAL archives,

work page

[2] [11]

Exploring Conceptual Modeling Metaphysics: Existence Containers, Leibniz's Monads and Avicenna's Essence

S. Al-Fedaghi, “Exploring conceptual modeling metaphysics: existence containers, Leibniz’s monads and Avicenna’s essence,” arXiv preprint, arXiv:2405.01549, February 20, 2024

work page Pith review arXiv 2024

[3] [1]

Conceptual modeling supported by text analysis,

B. Otsuka, “Conceptual modeling supported by text analysis,” Procedia Computer Science, vol. 126, pp. 1387-1394, 2018

work page 2018

[4] [2]

Text-based modeling,

H. Gronniger, H. Krahn, B. Rumpe, M. Schindler, and S. Volkel, “Text-based modeling,” 4th International Workshop on Software Language Engineering (ateM 2007), Nashville, TN, USA, October 2007. https://www.researchgate.net/publication/266024095_Textbased _Modeling

work page arXiv 2007

[5] [3]

Automated domain modeling with large language models: a comparative study,

K. Chen, Y. Yang, B. Chen, J. A. Hernandez Lopez, G. Mussbacher, and D. Varro, “Automated domain modeling with large language models: a comparative study,” ACM/IEEE 26th International Conference on Model Driven Engineering Languages and Systems (MODELS), Västerås, Sweden. October 1-6, 2023

work page 2023

[6] [4]

Framework for conceptual modeling on natural language texts, 13-24,

M. Bogatyrev and K. Samodurov, “Framework for conceptual modeling on natural language texts, 13-24,” Third Workshop on Concept Discovery in Unstructured Data, Moscow, Russia, July 18, 2016. https://ceur-ws.org/Vol-1625/paper2.pdf

work page 2016

[7] [5]

Leveraging LLMs in text-based ontology- driven conceptual modeling,

M. L. Coutinho, “Leveraging LLMs in text-based ontology- driven conceptual modeling,” July 15, 2024 - September 19, 2024, Enschede, Netherlands. https://www.utwente.nl/en/eemcs/fois2024/resources/papers/cout inho-leveraging-llms-in-text-based-ontology-driven-conceptual- modeling.pdf

work page 2024

[8] [6]

Text-to-model transformation: natural language-based model generation framework

A. Akundi, J. Ontiveros, and S. Luna, “Text-to-model transformation: natural language-based model generation framework.” Systems, vol. 12, no. 369. https://doi.org/10.3390/systems12090369

doi:10.3390/systems12090369