Marketing a Banned Remedy: A Topic Model Analysis of Health Misinformation in Thai E-Commerce

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study applies Latent Dirichlet Allocation (LDA; a common topic modeling approach) to examine the marketing discourse surrounding a unregulated herbal product on Thai digital media platforms. The integration of NLP techniques with public health aligns well with the journal's scope and the section's focus. The mixed-methods design and the use of both topic modeling are rigorous. In addition, Health Belief Model provides theoretical foundation of this study, bridging LDA with health behavior theory. Minor issues include that probabilistic-based LDA might not be the most up to date topic modeling method compared to other deep learning based alternatives. Topic labeling could be further explained (e.g., how inter-rater reliability is assured). Discussions on scientific contributions (rather than adopting existing methodology) should be noted. 

Author Response

To the Reviewer,

We would like to express our sincere thanks for your insightful feedback and for recognizing the rigorous design of our study. We appreciate your valuable suggestions on how to further strengthen the manuscript and have carefully addressed each point.

For your convenience, all additions and significant revisions made to the manuscript have been highlighted in green text.

Below is a summary of the key revisions made in response to your comments:

• 1. Justification of LDA Methodology: Regarding your comment on the choice of LDA compared to newer deep learning alternatives, we have added a detailed justification to the 'Methodological Reflections' section (5.6). This new text clarifies that while newer models exist, LDA was deliberately selected for its high degree of interpretability, which was essential for our primary goal of linking the computational findings to the study's theoretical frameworks (the Health Belief Model and Information Manipulation Theory).
• 2. Clarification of the Topic Labeling Process: To address your suggestion for a clearer explanation of the topic labeling process, we have expanded the 'Analytical Procedures' section (3.5). This section now includes a detailed description of the systematic, multi-step method used. We explain how a consensus-based approach, involving two researchers who independently reviewed keywords and documents, was employed to ensure the reliability and validity of the final topic labels.
• 3. Highlighting Scientific Contributions: Per your valuable recommendation, we have made our study's scientific contributions more explicit. We have added a dedicated paragraph to the 'Conclusion' (Section 6) that clearly articulates the study's specific methodological, theoretical, and empirical contributions to the fields of public health, digital governance, and computational social science.

We believe these revisions directly address your feedback and have significantly improved the manuscript's clarity and scholarly contribution. Thank you once again for your time and constructive guidance.

Sincerely,

Dr. Kanitsorn Suriyapaiboonwattana

Reviewer 2 Report

Comments and Suggestions for Authors

This article is highly relevant, innovative, and well-structured. The author(s) successfully address an urgent and globally significant issue—the spread of health misinformation in digital environments. The use of Latent Dirichlet Allocation (LDA) to analyze the marketing narratives and manipulative tactics surrounding Jindamanee powder represents a methodological advance in research on digital commerce and public health. The paper carries strong academic value as well as practical relevance for regulators and policymakers. The integration of Information Manipulation Theory (IMT) and the Health Belief Model (HBM) is well-reasoned and highly effective. I especially appreciate how these frameworks are not merely cited but meaningfully applied to guide coding and interpretive analysis. The mixed-methods design—combining descriptive statistics and LDA topic modeling—demonstrates a high level of research sophistication. The preprocessing steps (emoji removal, translation, character normalization) are clearly described and contribute to the transparency and reproducibility of the study. The article is logically organized into coherent sections: contextual background, theoretical framework, methodology, analysis, and conclusions. Each part is purposeful and contributes meaningfully to the overall narrative. The insights into misinformation architecture, semantic patterns, and marketing strategies on platforms like Facebook and TikTok are not only academically valuable but also practically useful for public health authorities. The finding that 87% of listings appear on Facebook—despite its policies—is particularly compelling. 

Areas for Improvement 

Title Refinement

While accurate, the title could be slightly more descriptive. I suggest something like:
“Digital Health Misinformation in Thai E-Commerce: A Topic Model Analysis of Jindamanee Powder Promotion on Social Platforms”

Limitations section

The limitations are implied (e.g., translation noise, content filtering), but it would be helpful to include a dedicated paragraph outlining:
• Sampling limitations
• Potential translation inaccuracies
• Lack of longitudinal analysis (the study is based on a cross-sectional snapshot)

Clarification of Product Definition and Regulatory Status

While the paper assumes prior familiarity with Jindamanee powder, international readers may benefit from a brief but explicit explanation of what the product is, its traditional uses, and why it is banned. Including this information early in the introduction would clarify the significance of the regulatory violations discussed later in the paper and better frame the public health implications.

Stronger Theoretical Integration in the Discussion

Although IMT and HBM are well-applied throughout, I would recommend a slightly deeper synthesis in the discussion section—particularly by mapping findings more explicitly to HBM components like “perceived severity,” “barriers,” and “benefits.”

Writing Style

While the academic tone is appropriate, some parts of the text have a mechanical cadence (e.g., frequent repetition of phrases like “this study revealed” or “Topic X showed that”). Slight variation in sentence structure would improve fluency and give the paper a more “human” voice.

Comments on the Quality of English Language

The manuscript is generally well-written, clear, and grammatically sound. While I am not a native speaker nor an expert in English language editing, I found the academic tone appropriate and the argumentation coherent. Nonetheless, I strongly recommend a professional language review to ensure fluency, stylistic consistency, and the elimination of any residual mechanical phrasing.

Author Response

To the Reviewer 2,

We would like to express our sincere gratitude for your detailed and highly constructive feedback on our manuscript. Your insights have been instrumental in improving the paper's rigour, clarity, and overall impact. We have carefully addressed each of your suggestions in the revised version.

For your convenience, all additions and significant revisions made to the manuscript have been highlighted in green text.

Below is a point-by-point summary of the key revisions made in response to your comments:

• 1. Title Refinement: As recommended, we have revised the title to be more descriptive and compelling. The new title is: "Marketing a Banned Remedy: A Topic Model Analysis of Health Misinformation in Thai E-Commerce." We chose this option because the main title serves as an effective narrative hook, while the subtitle clearly outlines the methodology, core theme, and context, directly addressing your suggestion for a more descriptive title.
• 2. Clarification of Product Definition and Status: To provide essential context for an international audience, we have significantly revised the Introduction (Section 1). The original third paragraph has been replaced with a more comprehensive one that synthesizes the product's purported traditional use (for pain and inflammation) with the specific reasons for its ban (adulteration with named substances like dexamethasone). This revision creates a clearer narrative for readers who may be unfamiliar with the case study.
• 3. Stronger Theoretical Integration in the Discussion: We agree that a deeper synthesis with the Health Belief Model (HBM) strengthens the paper. We have therefore expanded Section 5.5 (Theoretical Implications) to explicitly link the empirical findings from our topic model to the HBM's core constructs. For example, we now demonstrate how sellers amplify "perceived severity" by focusing on topics related to chronic pain (e.g., Topic 2, 5), while simultaneously lowering "perceived barriers" through topics centered on discounts and affordability (e.g., Topic 1, 4). This explicit connection provides the deeper theoretical analysis you requested.
• 4. Dedicated Limitations Section: Following your guidance, we have restructured Section 5.8 to create a more explicit "Limitations and Future Research Directions" section. The revised text is now organized more formally, beginning with the temporal and sampling limitations (our study being a "snapshot in time"), followed by methodological constraints (including potential translation inaccuracies), and concluding with the theoretical and compositional boundaries of our work. This structure makes the scope of our study transparent and offers clearer pathways for future research.
• 5. Writing Style and Fluency: We have thoroughly revised the entire manuscript to improve the writing style and address the "mechanical cadence" you noted. Passive sentence constructions (e.g., "The analysis revealed...") have been rephrased into more active forms (e.g., "A notable consistency emerged...") throughout the paper to enhance fluency. We believe these changes, combined with a final professional language review, have given the paper a more "human voice" and significantly improved its readability.

We are confident that these revisions have substantially improved the manuscript. Thank you once again for your time and expert guidance, which has helped us to elevate the quality of our work
Best Regards,

Dr. Kanitsorn Suriyapaiboonwattana