pith. sign in

arxiv: 2605.02384 · v1 · submitted 2026-05-04 · 💻 cs.SE · cs.HC

A Low-Code Approach for the Automatic Personalization of Conversational Agents

Aaron Conrardy , Alfredo Capozucca , Jordi Cabot This is my paper

Pith reviewed 2026-05-08 18:09 UTC · model grok-4.3

classification 💻 cs.SE cs.HC
keywords user modelingmodel-driven engineeringMDEsystematic literature reviewpersonalizationuser profilesadaptationroadmap
0
0 comments X

The pith

User modeling in model-driven engineering consists of disconnected proposals covering only partial and mostly static dimensions with limited tool support.

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

The paper conducts a systematic literature review to assess user modeling in the model-driven engineering domain. It finds existing proposals to be diverse but disconnected, covering only a partial set of dimensions while emphasizing easier-to-profile characteristics and treating most as fixed rather than dynamically evolving with user interactions. Tool support remains narrow and focused on initial model creation. The authors propose a roadmap calling for community agreement on a unified reusable user model, machine learning methods for automatic incremental profile derivation from interactions, and automated pipelines to transform profiles into application adaptations that personalize software to user needs.

Core claim

Results show a diverse set of disconnected proposals, covering a partial number of dimensions with an emphasis on those characteristics that are easier to profile. Moreover, most dimensions are regarded as fixed instead of allowing their dynamic evolution during the interaction with the software application. Tool support is rather limited, mostly restricted to enabling the creation of the user models itself. The roadmap calls for a unified and re-usable user model covering the superset of all dimensions plus others from domains such as sociology, ML-based proposals to automatically and incrementally derive user profiles from interactions, and automatic pipelines to transform user information

What carries the argument

systematic literature review that maps proposals by dimension coverage, dynamism and tooling to identify gaps and motivate a unification roadmap

If this is right

  • Community consensus on a unified user model would combine the full range of dimensions from existing literature with additions from other fields.
  • ML-based techniques would automatically derive and update user profiles incrementally from interaction data.
  • Automatic pipelines would convert user profiles into concrete adaptations that personalize applications to individual needs.
  • Allowing dimensions to evolve dynamically would support ongoing adaptation during software use rather than one-time profiling.

Where Pith is reading between the lines

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

  • The low-code emphasis in the paper title suggests the roadmap could be delivered through model-driven tools that abstract profile management and adaptation logic.
  • Applying the automatic pipelines to conversational agents would enable runtime personalization of responses and flows based on evolving user profiles.
  • Incorporating sociological dimensions into the unified model could reveal user traits overlooked in current MDE work.

Load-bearing premise

The systematic literature review is comprehensive and the identified gaps accurately represent the field so that the roadmap directions can be advanced without further validation.

What would settle it

A new survey locating many integrated, dynamic user modeling proposals with robust tooling or an implementation of the ML pipelines and unified model showing no personalization gains would challenge the findings.

Figures

Figures reproduced from arXiv: 2605.02384 by Aaron Conrardy, Alfredo Capozucca, Jordi Cabot.

Figure 1
Figure 1. Figure 1: Overview of steps of low-code personalized agent creation pipeline view at source ↗
Figure 2
Figure 2. Figure 2: High-level view of user metamodel from [9] view at source ↗
Figure 3
Figure 3. Figure 3: Paraplegic user profile model in graphical notation view at source ↗
Figure 4
Figure 4. Figure 4: Agent and agent profile metamodel view at source ↗
Figure 5
Figure 5. Figure 5: Graphical agent model representing the gym assistant agent view at source ↗
Figure 6
Figure 6. Figure 6: Form-based configuration specification 3.4 Linking the models In a final step, the personalization model needs to be mapped to the affected base agent model and target user profile, illustrated in view at source ↗
Figure 7
Figure 7. Figure 7: Prompt for concise sentence generation view at source ↗
Figure 8
Figure 8. Figure 8: Process for nondeterministic personalization aspects view at source ↗
Figure 9
Figure 9. Figure 9: Profile picker and chat with personalized agent for paraplegic user view at source ↗
read the original abstract

In this paper, we conducted an SLR on the state of user modeling in the MDE domain. Results show a diverse set of disconnected proposals, covering a partial number of dimensions with an emphasis on those characteristics that are easier to profile. Moreover, most dimensions are regarded as fixed instead of allowing their dynamic evolution during the interaction with the software application. It is also worth noting that tool support is also rather limited, mostly limited to enabling the creation of the user models itself. The roadmap we hope to see in this area stems from the discussion points seen above. For instance, we believe the community should agree on a unified and re-usable user model, covering the superset of all dimensions present in the literature. Plus additional ones we could learn from user profiling in other domains (e.g. sociology). On the technical side, we expect to see a new generation of ML-based proposals to automatically and incrementally derive a user profile from the analysis of user interactions and a number of automatic pipelines able to transform the user information in concrete application adaptations that personalize the application to cater to the user's needs and profile.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 1 minor

Summary. The manuscript presents the results of a systematic literature review (SLR) on user modeling in the Model-Driven Engineering (MDE) domain. Key findings include a diverse set of disconnected proposals that cover only a partial number of user dimensions, with emphasis on easily profileable characteristics; most dimensions are treated as fixed rather than dynamically evolving during user interactions; and tool support is limited, mainly to user model creation. The paper concludes by proposing a roadmap for the field, advocating for a unified and reusable user model, ML-based incremental profiling from user interactions, and automatic pipelines for application adaptations to personalize based on user profiles.

Significance. If the SLR is methodologically sound, the findings on fragmentation and lack of dynamic modeling in MDE user modeling could be significant for advancing personalized software engineering. The roadmap identifies actionable directions that could lead to more adaptive systems, particularly if combined with low-code techniques. However, the mismatch with the title's promise of a concrete low-code approach for conversational agents reduces the paper's immediate contribution, as no such method is developed or even outlined in detail. The work has potential to influence the MDE and personalization communities by highlighting gaps, but requires stronger empirical grounding for the proposed future directions.

major comments (3)
  1. Title: the title promises 'A Low-Code Approach for the Automatic Personalization of Conversational Agents', yet the manuscript contains no description of a low-code approach, no specific techniques for conversational agents, and no implementation or pipeline; the content is limited to an SLR on MDE user modeling and a high-level roadmap. This discrepancy means the central claim of the title is unsupported by the delivered work.
  2. Abstract: the abstract reports SLR outcomes on disconnected proposals, fixed dimensions, and limited tool support but provides no details on the search strategy, inclusion criteria, data extraction, or quality assessment methods. Without these, the robustness of the findings cannot be verified, as required for a credible SLR.
  3. Roadmap section: the proposed roadmap suggests ML-based incremental profiling and automatic adaptation pipelines but offers no concrete mechanisms, examples, or validation steps, nor does it address how these would constitute a low-code approach. This leaves the suggestions at a high level without evidence of feasibility.
minor comments (1)
  1. Introduction: terms such as 'MDE domain' and the specific 'dimensions' of user models could be defined more explicitly early in the paper to aid readers from outside the subfield.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We acknowledge the valid concerns regarding the title's mismatch with the manuscript content, the absence of SLR methodology details in the abstract, and the high-level presentation of the roadmap. We will revise the manuscript to address these points directly while maintaining the integrity of the SLR findings and the proposed directions.

read point-by-point responses

  1. Referee: Title: the title promises 'A Low-Code Approach for the Automatic Personalization of Conversational Agents', yet the manuscript contains no description of a low-code approach, no specific techniques for conversational agents, and no implementation or pipeline; the content is limited to an SLR on MDE user modeling and a high-level roadmap. This discrepancy means the central claim of the title is unsupported by the delivered work.

    Authors: We agree that the title does not accurately represent the delivered work. The manuscript focuses on an SLR of user modeling in MDE and a high-level roadmap, without developing or outlining any concrete low-code approach or specific techniques for conversational agents. The title was selected to emphasize the broader motivation and potential application area, but this creates a misleading impression. We will revise the title to something such as 'User Modeling in Model-Driven Engineering: A Systematic Literature Review and Roadmap for Low-Code Personalization' to align precisely with the content. revision: yes

  2. Referee: Abstract: the abstract reports SLR outcomes on disconnected proposals, fixed dimensions, and limited tool support but provides no details on the search strategy, inclusion criteria, data extraction, or quality assessment methods. Without these, the robustness of the findings cannot be verified, as required for a credible SLR.

    Authors: The observation is correct; the abstract as currently written omits essential methodological information required to assess an SLR. We will expand the abstract to include a brief but explicit description of the search strategy (databases and keywords used), inclusion and exclusion criteria, the data extraction process, and the quality assessment methods applied to the selected studies. revision: yes

  3. Referee: Roadmap section: the proposed roadmap suggests ML-based incremental profiling and automatic adaptation pipelines but offers no concrete mechanisms, examples, or validation steps, nor does it address how these would constitute a low-code approach. This leaves the suggestions at a high level without evidence of feasibility.

    Authors: We recognize that the roadmap remains conceptual and does not provide concrete mechanisms, examples, or explicit links to low-code techniques. As the paper's primary contribution is the SLR synthesis and gap identification, the roadmap was intentionally kept at a strategic level to guide future research. To improve this section, we will add illustrative examples of how ML-driven incremental profiling could integrate with MDE tools and discuss potential low-code platforms that might support adaptation pipelines, while clarifying that detailed validation and implementation fall outside the scope of this review. revision: partial

Circularity Check

0 steps flagged

No circularity: descriptive SLR with roadmap, no derivations or self-referential reductions

full rationale

The paper conducts and reports an SLR on user modeling in MDE, summarizes findings on disconnected proposals and fixed dimensions, and outlines a high-level future roadmap (unified model, ML-based profiling, adaptation pipelines). No equations, algorithms, fitted parameters, predictions, or derivation chains exist. No self-citations are invoked as load-bearing premises, and the central claims are observational summaries rather than constructed outputs. The content is self-contained as a review without any step that reduces to its own inputs by definition or construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper is a literature review relying on standard SLR methodology assumptions without new parameters, axioms, or invented entities.

pith-pipeline@v0.9.0 · 8913 in / 962 out tokens · 51073 ms · 2026-05-08T18:09:24.130417+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

  1. [1]

    The power of personalization: A systematic review of personality-adaptive chatbots.SN Computer Science, 4(5), 2023

    Tarek Ait Baha, Mohamed El Hajji, Youssef Es-Saady, and Hammou Fadili. The power of personalization: A systematic review of personality-adaptive chatbots.SN Computer Science, 4(5), 2023. doi:10.1007/ s42979-023-02092-6

    work page 2023

  2. [2]

    Building besser: an open-source low-code platform

    Iván Alfonso, Aaron Conrardy, Armen Sulejmani, Atefeh Nirumand, Fitash Ul Haq, Marcos Gomez-Vazquez, et al. Building besser: an open-source low-code platform. InInternational Conference on Business Process Modeling, Development and Support, pages 203–212, 2024.doi:10.1007/978-3-031-61007-3_16

    work page doi:10.1007/978-3-031-61007-3_16 2024

  3. [3]

    Mohammad Amin Kuhail, Mohamed Bahja, Ons Al-Shamaileh, Justin Thomas, Amina Alkazemi, and Joao Negreiros. Assessing the impact of chatbot-human personality congruence on user behavior: A chatbot-based advising system case.IEEE Access, 12:71761–71782, 2024.doi:10.1109/ACCESS.2024.3402977

    work page doi:10.1109/access.2024.3402977 2024

  4. [4]

    Cutrupi, Maristella Matera, Isabella Possaghi, Emanuele Pucci, et al

    Marcos Baez, Cinzia Cappiello, Claudia M. Cutrupi, Maristella Matera, Isabella Possaghi, Emanuele Pucci, et al. Supporting natural language interaction with the web. InWeb Engineering: 22nd International Conference, ICWE 2022, Bari, Italy, July 5–8, 2022, Proceedings, page 383–390, 2022.doi:10.1007/978-3-031-09917-5_26

    work page doi:10.1007/978-3-031-09917-5_26 2022

  5. [5]

    Lamya Benaddi, Charaf Ouaddi, Abdeslam Jakimi, Hasna Chaibi, Abdellah Chehri, Gwanggil Jeon, et al. A survey on model-driven engineering and domain-specific languages for chatbot development: Requirements, challenges and solutions.Expert Systems, 42(10):e70124, 2025.doi:10.1111/exsy.70124

    work page doi:10.1111/exsy.70124 2025

  6. [6]

    Morgan Kaufmann, 2014

    Marco Brambilla and Piero Fraternali.Interaction flow modeling language: Model-driven UI engineering of web and mobile apps with IFML. Morgan Kaufmann, 2014

    work page 2014

  7. [7]

    quick and dirty

    John Brooke. Sus: A “quick and dirty” usability scale. InUsability Evaluation in Industry, pages 189–194. 1996

    work page 1996

  8. [8]

    Low code for smart software development.IEEE Software, 40(1):89–93, 2023

    Jordi Cabot and Robert Clarisó. Low code for smart software development.IEEE Software, 40(1):89–93, 2023. doi:10.1109/MS.2022.3211352

    work page doi:10.1109/ms.2022.3211352 2023

  9. [9]

    Towards a unified user modeling language for engineering human centered ai systems, 2025

    Aaron Conrardy, Alfredo Capozucca, and Jordi Cabot. Towards a unified user modeling language for engineering human centered ai systems, 2025. URL:https://arxiv.org/abs/2505.24697

    work page arXiv 2025

  10. [10]

    User modeling in model-driven engineering: A systematic literature review.Journal of Object Technology, 24(2):2:1–14, 2025.doi:10.5381/jot.2025.24.2.a12

    Aaron Conrardy, Alfredo Capozucca, and Jordi Cabot. User modeling in model-driven engineering: A systematic literature review.Journal of Object Technology, 24(2):2:1–14, 2025.doi:10.5381/jot.2025.24.2.a12

    work page doi:10.5381/jot.2025.24.2.a12 2025

  11. [11]

    Fred D. Davis. Perceived usefulness, perceived ease of use, and user acceptance of information technology.MIS Quarterly, 13(3):319–340, 1989.doi:10.2307/249008

    work page doi:10.2307/249008 1989

  12. [12]

    Low-code development and model-driven engineering: Two sides of the same coin?Software and Systems Modeling, 21:437–446, 2022.doi:10.1007/s10270-021-00970-2

    Davide Di Ruscio, Dimitris Kolovos, Juan Lara, Alfonso Pierantonio, Massimo Tisi, and Manuel Wimmer. Low-code development and model-driven engineering: Two sides of the same coin?Software and Systems Modeling, 21:437–446, 2022.doi:10.1007/s10270-021-00970-2

    work page doi:10.1007/s10270-021-00970-2 2022

  13. [13]

    Haiyan Fan and Marshall Scott Poole. What is personalization? perspectives on the design and implementation of personalization in information systems.Journal of Organizational Computing and Electronic Commerce, 16(3-4):179–202, 2006.doi:10.1080/10919392.2006.9681199. 11 Low-Code Personalization of Conversational Agents

    work page doi:10.1080/10919392.2006.9681199 2006

  14. [14]

    International Journal of Human-Computer Studies106, 63–76 (2017) https://doi.org/10.1016/j.ijhcs

    Jasper Feine, Ulrich Gnewuch, Stefan Morana, and Alexander Maedche. A taxonomy of social cues for conversa- tional agents.International Journal of Human-Computer Studies, 132:138–161, 2019. doi:10.1016/j.ijhcs. 2019.07.009

    work page doi:10.1016/j.ijhcs 2019

  15. [15]

    The Internet Movie Database Case Study, pages 65–108

    Joan Fons, Vicente Pelechano, Oscar Pastor, Pedro Valderas, and Victoria Torres.Applying the Oows Model-Driven Approach for Developing Web Applications. The Internet Movie Database Case Study, pages 65–108. Springer London, 2008.doi:10.1007/978-1-84628-923-1_5

    work page doi:10.1007/978-1-84628-923-1_5 2008

  16. [16]

    Syed Mahmudul Huq, Rytis Maskeli¯unas, and Robertas Damaševiˇcius. Dialogue agents for artificial intelligence- based conversational systems for cognitively disabled: a systematic review.Disability and Rehabilitation: Assistive Technology, 19(3):1059–1078, 2024.doi:10.1080/17483107.2022.2146768

    work page doi:10.1080/17483107.2022.2146768 2024

  17. [17]

    Quiroz, Liliana Laranjo, Huong Ly Tong, Darius Rezazadegan, Andrea Briatore, and Enrico Coiera

    Ahmet Baki Kocaballi, Shlomo Berkovsky, Juan C Quiroz, Liliana Laranjo, Huong Ly Tong, Dana Rezazadegan, et al. The personalization of conversational agents in health care: Systematic review.J Med Internet Res, 21(11):e15360, 2019.doi:10.2196/15360

    work page doi:10.2196/15360 2019

  18. [18]

    Springer London, 2008.doi:10.1007/978-1-84628-923-1_7

    Nora Koch, Alexander Knapp, Gefei Zhang, and Hubert Baumeister.Uml-Based Web Engineering, pages 157–191. Springer London, 2008.doi:10.1007/978-1-84628-923-1_7

    work page doi:10.1007/978-1-84628-923-1_7 2008

  19. [19]

    Culturellm: Incorporating cultural differences into large language models

    Cheng Li, Mengzhuo Chen, Jindong Wang, Sunayana Sitaram, and Xing Xie. Culturellm: Incorporating cultural differences into large language models. InAdvances in Neural Information Processing Systems, volume 37, pages 84799–84838, 2024.doi:10.52202/079017-2693

    work page doi:10.52202/079017-2693 2024

  20. [20]

    Emiliana Murgia, Maria Soledad Pera, Monica Landoni, and Theo Huibers. Children on chatgpt readability in an educational context: Myth or opportunity? InAdjunct Proceedings of the 31st ACM Conference on User Modeling, Adaptation and Personalization, page 311–316, 2023.doi:10.1145/3563359.3596996

    work page doi:10.1145/3563359.3596996 2023

  21. [21]

    Nagarhalli, Vinod Vaze, and N

    Tatwadarshi P. Nagarhalli, Vinod Vaze, and N. K. Rana. A review of current trends in the development of chatbot systems. In2020 6th International Conference on Advanced Computing and Communication Systems (ICACCS), pages 706–710, 2020.doi:10.1109/ICACCS48705.2020.9074420

    work page doi:10.1109/icaccs48705.2020.9074420 2020

  22. [22]

    The art of talking machines: A comprehensive literature review of conversational user interfaces

    Mohammad (Matt) Namvarpour and Afsaneh Razi. The art of talking machines: A comprehensive literature review of conversational user interfaces. InProceedings of the 7th ACM Conference on Conversational User Interfaces, 2025.doi:10.1145/3719160.3736621

    work page doi:10.1145/3719160.3736621 2025

  23. [23]

    Generating voice user interfaces from web sites.Behaviour & Information Technology, 43(13):3129– 3152, 2024.doi:10.1080/0144929X.2023.2272192

    Gonzalo Ripa, Manuel Torre, Matias Urbieta, Gustavo Rossi, Alejandro Fernandez, Alex Tacuri, and Sergio Firmenich. Generating voice user interfaces from web sites.Behaviour & Information Technology, 43(13):3129– 3152, 2024.doi:10.1080/0144929X.2023.2272192

    work page doi:10.1080/0144929x.2023.2272192 2024

  24. [24]

    Chatbot design and implementation: Towards an operational model for chatbots.Information, 15(4), 2024.doi:10.3390/info15040226

    Alexander Skuridin and Martin Wynn. Chatbot design and implementation: Towards an operational model for chatbots.Information, 15(4), 2024.doi:10.3390/info15040226

    work page doi:10.3390/info15040226 2024

  25. [25]

    Adaptive and accessible user interfaces for seniors through model-driven engineering.Automated Software Engineering, 32(2):74, 2025

    Shavindra Wickramathilaka, John Grundy, Kashumi Madampe, and Omar Haggag. Adaptive and accessible user interfaces for seniors through model-driven engineering.Automated Software Engineering, 32(2):74, 2025. doi:10.1007/s10515-025-00547-z

    work page doi:10.1007/s10515-025-00547-z 2025

  26. [26]

    Integrated model-driven development of self-adaptive user interfaces.Softw

    Enes Yigitbas, Ivan Jovanovikj, Kai Biermeier, Stefan Sauer, and Gregor Engels. Integrated model-driven development of self-adaptive user interfaces.Softw. Syst. Model., 19(5):1057–1081, 2020. doi:10.1007/ s10270-020-00777-7

    work page 2020

  27. [27]

    Multilingual Machine Translation with Large Language Models: Empirical Results and Analysis , booktitle =

    Wenhao Zhu, Hongyi Liu, Qingxiu Dong, Jingjing Xu, Shujian Huang, Lingpeng Kong, et al. Multilingual machine translation with large language models: Empirical results and analysis. InFindings of the Association for Compu- tational Linguistics: NAACL 2024, pages 2765–2781, 2024. doi:10.18653/v1/2024.findings-naacl.176. 12

    work page doi:10.18653/v1/2024.findings-naacl.176 2024