Investigation of Corporate Sustainability Performance Data and Developing an Innovation-Oriented Novel Analysis Method with Multi-Criteria Decision Making Approach

Response to Reviewer-1 Comments

1. Summary

We would like to express our sincere gratitude for the time and effort you devoted to reviewing our manuscript. Your constructive comments and insightful recommendations have been invaluable in strengthening both the quality and presentation of our work. In response, we have undertaken several key revisions. Specifically, we elaborated the Introduction to provide clearer context and research motivation, refined the Methodology section with more transparent descriptions of data collection and analysis procedures, and expanded the Results and Discussion with additional comparisons to recent studies in the field. We also improved the readability of the manuscript by restructuring some paragraphs, enriching the figures and tables for better clarity, and updating the reference list to include recent and relevant contributions. We trust that these revisions have enhanced the rigor, coherence, and contribution of the study.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Is the content succinctly described and contextualized with respect to previous and present theoretical background and empirical research (if applicable) on the topic?

Can be improved

Thank you for your valuable comment. In response, we have revised Sections 1 and 2 to strengthen the theoretical background and to provide a clearer link between prior research and the present study. We believe these revisions improve the contextualization and address your concern.

Are the research design, questions, hypotheses and methods clearly stated?

Can be improved

Thank you for this important comment. In response, we have revised manuscript to more clearly present the research design and methodological framework. We explicitly stated the research questions and formulated hypotheses to guide the analysis, thereby strengthening the link between theory and empirical investigation. Furthermore, we elaborated on the use of DEMATEL, the ANN-based reverse DEMATEL model, and the Entropy method, along with the expert evaluation process, to ensure methodological transparency. We believe these revisions improve the clarity and rigor of the study’s design, questions, hypotheses, and methods.

Are the arguments and discussion of findings coherent, balanced and compelling?

Must be improved

Thank you for this valuable comment. In response, we have substantially revised the Discussion section to improve the coherence and persuasiveness of our arguments. The section is now more clearly structured, with findings explicitly linked to both theoretical perspectives and recent empirical studies in sustainability research. We also balanced the discussion by acknowledging potential limitations and outlining implications for both academia and practice. In addition, a dedicated subsection on future research directions has been included to ensure a more comprehensive treatment of the findings. We believe these improvements have enhanced the overall clarity, balance, and impact of the discussion.

For empirical research, are the results clearly presented?

Can be improved

Thank you for this helpful comment. We have revised the Results to present the empirical findings more clearly and transparently. First, manuscript is reorganized and we integrated sensitivity/robustness routines into the analysis software and documented them in Appendix B: k-fold cross-validation for generalizability; architectural sensitivity (layers/neurons/activations); noise-robustness tests via Gaussian perturbations; local sensitivity via Jacobian norms; and stability checks on direct-relation matrices via Frobenius norms. We also report MSE and MAE to summarize predictive error and include compact plots summarizing variability across folds and perturbation levels. We believe these additions make the empirical results clearer, reproducible, and easier to interpret.

Is the article adequately referenced?

Can be improved

Thank you for this important comment. In response, we have substantially expanded and updated the reference list to ensure comprehensive coverage of both classical and recent works in sustainability and multi-criteria decision-making. Several new references from the Sustainability journal and other high-impact outlets have been added to strengthen the theoretical foundation and contextualize our findings within the current literature. In addition, company reports were formatted in line with MDPI standards to ensure consistency and accuracy. We believe these improvements have significantly enhanced the quality, relevance, and adequacy of the referencing throughout the manuscript.

Are the conclusions thoroughly supported by the results presented in the article or referenced in secondary literature?

Must be improved

Thank you for this valuable comment. In response, we have revised the Conclusions section to ensure that the main points are clearly supported by the results and consistent with the discussion. The conclusions are now explicitly linked to the research questions and hypotheses formulated in the study, showing how the findings address and support them. In addition, we incorporated relevant literature to reinforce our interpretations and added implications for practice and future research. We believe these revisions make the conclusions clearer, more coherent, and convincingly grounded in both our results and the broader scholarly context.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1:

The exclusive focus on Turkish firms (BIST Sustainability Index) limits cross-cultural applicability. Findings may not extend to markets with differing regulatory or economic contexts.

Response 1:

We sincerely thank the reviewer for this important observation. We agree that cross-cultural applicability is an essential consideration in sustainability research. In revising the manuscript, we carefully clarified why the exclusive focus on Turkish firms listed on the Borsa Istanbul (BIST) Sustainability Index does not undermine, but rather reinforces, the international relevance of the study.

1.       International Orientation of the BIST Sustainability Index

o    The BIST Sustainability Index is not an isolated, locally designed benchmark. It was developed in collaboration with Borsa Istanbul and Ethical Investment Research Services (EIRIS, London, now part of Moody’s ESG Solutions), aligning directly with global ESG evaluation frameworks.

o    The index methodology explicitly references internationally recognized standards such as the UN Global Compact Principles, the OECD Guidelines for Multinational Enterprises, the OECD Principles of Corporate Governance, and the International Labour Organization (ILO) conventions.

o    As such, companies included in the BIST Sustainability Index are evaluated using globally harmonized criteria, ensuring comparability with sustainability indices in other markets (e.g., Dow Jones Sustainability Index, FTSE4Good).

2.       Global Character of the Selected Firms

o    The ten firms analyzed in this study—such as Arçelik (ARCLK), Ford Otosan (FROTO), Enerjisa (ENJSA), and Türkcell (TCELL)—are not only leading companies in Türkiye but also globally active corporations with joint ventures, subsidiaries, and partnerships extending across multiple continents.

o    For instance, Arçelik operates over 20 R&D centers in Europe, Asia, and Africa, and its sustainability strategies are benchmarked against EU environmental regulations.

o    Ford Otosan functions as part of Ford Motor Company’s global production network, directly aligning with the parent firm’s international ESG and innovation strategies.

o    Enerjisa, jointly owned by E.ON (Germany) and Sabancı Holding, reflects a hybrid corporate governance structure embedded in European sustainability regulations.

o    Türkcell, as a telecommunications giant, is listed on the New York Stock Exchange (NYSE) and complies with SEC disclosure requirements, ensuring global accountability.

o    Similar cases apply to other firms (e.g., Şişecam, a global glass manufacturer exporting to over 150 countries, and Ülker, part of Yıldız Holding with international brands like Godiva).

3.       Cross-Sectoral and Cross-Market Comparability

o    The firms cover a broad range of industries (chemicals, white goods, construction, energy, automotive, telecommunications, and food), which not only mirrors Türkiye’s industrial diversity but also facilitates comparison with global sectoral benchmarks.

o    This diversity ensures that the findings capture systemic dynamics of sustainability strategies across different regulatory pressures, environmental challenges, and financial structures—making the results relevant beyond the Turkish context.

4.       Ensuring International Relevance of Findings

o    While the dataset is drawn from Turkish companies, the methodological contribution—the integration of entropy with ANN-based reverse DEMATEL—is independent of geography and can be readily applied in other markets.

o    The cross-validation of results with international literature in the revised Discussion section further demonstrates that the patterns observed (e.g., R&D’s dual role in financial and environmental sustainability) are consistent with evidence reported in European, Asian, and North American studies.

In conclusion, although the study is based on Turkish firms, the selection of globally recognized, multinationally active, and internationally benchmarked corporations listed in the BIST Sustainability Index ensures that the insights are not only nationally contextualized but also globally relevant and transferable. The methodological framework, by design, is universal and suitable for application across different regulatory and cultural environments.

Comments 2:

While the ANN-based reverse DEMATEL is innovative, the absence of comparative validation against traditional DEMATEL (expert-driven) weakens reliability. Sensitivity analysis of the ANN architecture (e.g., layer choices) is also lacking.

Response 2:

We sincerely thank the reviewer for raising these important points concerning validation and sensitivity analysis. We recognize that any methodological contribution must be supported by rigorous tests to enhance reliability and scholarly confidence. In the revised manuscript, we addressed this comment in the following ways:

1. Comparative Validation with Traditional DEMATEL
• We explicitly discussed how the outputs of the proposed ANN-based reverse DEMATEL compare conceptually with those of the classical DEMATEL framework.
• As noted in the Materials and Methods and Discussion sections, the central purpose of the reverse approach is to reconstruct direct-relation matrices consistent with entropy-derived criterion weights, effectively simulating how expert-driven judgments would have been structured if the same weights were available.
• While a direct expert panel replication was not feasible within the scope of this study, we clarified that the structural patterns identified—particularly the prominence of R&D and greenhouse gas emissions—are consistent with findings in prior DEMATEL-based sustainability studies (e.g., Hsu et al., 2013; Tseng et al., 2019; Wang et al., 2020). This alignment provides an indirect validation of the reliability of our approach.
2. Sensitivity Analysis of ANN Architecture
• In the revised Materials and Methods section and Appendix B, we reported a series of robustness and sensitivity tests. These included:
  • Architectural Sensitivity: varying the number of hidden layers and neurons (2–4 layers, 32–128 neurons), where results showed consistent reconstruction accuracy.
  • Activation Functions: testing ReLU, Sigmoid, and Tanh, confirming that ReLU yielded the most stable outcomes.
  • Noise Robustness: introducing Gaussian noise to input weights and measuring changes using Frobenius norms, which indicated only minor deviations, confirming stability.
  • Local Sensitivity: analyzing Jacobian norms to assess how small perturbations in input weights affect the predicted direct-relation matrix, with results showing well-conditioned mappings.
• These analyses confirmed that the ANN architecture is not overly sensitive to minor design variations or data perturbations, reinforcing confidence in its robustness.
3. Positioning of Limitations
• In the newly added Limitations of the Work section, we acknowledged that the reconstructed interdependencies represent simulated approximations rather than definitive causal pathways. We emphasized, however, that this is not a weakness but a deliberate methodological innovation: it bridges the objectivity of entropy-based weights with the interpretability of causal mapping.
• We also clarified that the approach is complementary to, rather than a replacement for, traditional DEMATEL applications. Future research involving direct expert panels could further triangulate and strengthen validation.

In conclusion, through the integration of comparative discussion, extended robustness checks, and an explicit articulation of methodological scope, we have reinforced the reliability and practical applicability of the ANN-based reverse DEMATEL approach. We believe these revisions significantly enhance the rigor and transparency of the study.

Comments 3:

Figures referenced in Section 2.4 (e.g., "Impact Graphs") are not included, hindering reproducibility and clarity of bidirectional R&D-criteria interactions.

Response 3:

We sincerely thank the reviewer for this observation. In the revised manuscript, we clarified the placement and explanation of the figures to ensure transparency and readability. Specifically, the Impact Graphs are presented under Section 3.2, “Analyzing the Interaction Between Criteria, Centered on R&D Expenses.” Immediately after the subsection title, we added a general description of the visualization method:

“The visual representations derived from the direct-relation matrices of selected firms systematically depict the bidirectional interaction dynamics between environmental and financial sustainability indicators and R&D expenditures. The analysis employs a dual-axis DEMATEL-based framework, wherein the impact of each sustainability criterion on R&D expenditures is visualized through orange bars, while the reciprocal influence of R&D expenditures on these criteria is represented in yellow. This two-way interaction structure allows for a nuanced understanding of causal dependencies within the sustainability innovation nexus.”

Furthermore, to enhance clarity, each firm-specific figure is introduced with a clear reference (e.g., Figure 1, Figure 2, … Figure 10) at the beginning of its explanation, and the corresponding interpretation is provided directly below the figure. This structure ensures that readers can directly link each visualization with its explanatory analysis. By embedding both the figures and their detailed commentary within Section 3.2, we have strengthened reproducibility and improved the clarity of the bidirectional R&D-criteria interactions.

Comments 4:

Strategic recommendations are overly generic; sector-specific nuances (e.g., high-pollution vs. tech industries) are underexplored despite varied entropy weights.

Response 4:

We sincerely thank the reviewer for this insightful comment. In the revised manuscript, we addressed this issue by integrating sector-specific strategic recommendations directly into the interpretation of each firm’s Impact Graph (Figures 1–10). Specifically, at the end of each firm-level analysis, we added an additional paragraph that contextualizes the results within the firm’s sectoral characteristics and operational realities.

For example, for high-pollution and resource-intensive industries such as cement (CIMSA) and energy (ENJSA), the revised text highlights how R&D efforts should prioritize decarbonization, energy efficiency, and circular economy initiatives. By contrast, for technology-driven or consumer-oriented firms such as telecommunications (TCELL) and electronics/white goods (ARCLK), we emphasize innovation strategies that balance environmental responsibility with digital transformation and customer-focused sustainability. Similarly, in automotive (FROTO, TOASO) and glass/chemicals (SISE, AKSA, TKFEN), the discussion now explicitly considers how R&D must target sector-specific challenges such as lightweighting, alternative materials, and emissions control technologies. For food and beverage (ULKER), the recommendations now focus on sustainable sourcing, packaging innovation, and energy management within production systems.

Through these sectoral refinements, we ensured that the recommendations are no longer generic but instead reflect the heterogeneity of industries covered in the BIST Sustainability Index sample. This strengthens the managerial and policy relevance of our findings, demonstrates the applicability of the ANN-based reverse DEMATEL approach to varied industrial contexts, and provides more actionable insights for both practitioners and researchers.

4. Response to Comments on the Quality of English Language

Point 1:

The English is fine and does not require any improvement.

Response 1:

Thank you for your positive remark regarding the language quality of our manuscript. Following the overall revisions made throughout the paper, we have also made minor language refinements to ensure consistency and improve clarity. These edits were carried out with the support of a native English speaker to enhance fluency and readability. We believe these adjustments contribute positively to the overall presentation of the manuscript.

5. Additional clarifications

We would like to sincerely thank the reviewer and the editorial team for their insightful and constructive feedback. The comments provided have significantly contributed to the development and refinement of our manuscript. We have addressed all suggestions to the best of our ability, and we genuinely hope that the revised version meets your expectations.

Response to Reviewer 2 Comments

1. Summary

We would like to express our sincere gratitude for the time and effort you devoted to reviewing our manuscript. Your constructive comments and insightful recommendations have been invaluable in strengthening both the quality and presentation of our work. In response, we have undertaken several key revisions. Specifically, we elaborated the Introduction to provide clearer context and research motivation, refined the Methodology section with more transparent descriptions of data collection and analysis procedures, and expanded the Results and Discussion with additional comparisons to recent studies in the field. We also improved the readability of the manuscript by restructuring some paragraphs, enriching the figures and tables for better clarity, and updating the reference list to include recent and relevant contributions. We trust that these revisions have enhanced the rigor, coherence, and contribution of the study.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Is the content succinctly described and contextualized with respect to previous and present theoretical background and empirical research (if applicable) on the topic?

Can be improved

Thank you for your valuable comment. In response, we have revised Sections 1 and 2 to strengthen the theoretical background and to provide a clearer link between prior research and the present study. We believe these revisions improve the contextualization and address your concern.

Are the research design, questions, hypotheses and methods clearly stated?

Can be improved

Thank you for this important comment. In response, we have revised manuscript to more clearly present the research design and methodological framework. We explicitly stated the research questions and formulated hypotheses to guide the analysis, thereby strengthening the link between theory and empirical investigation. Furthermore, we elaborated on the use of DEMATEL, the ANN-based reverse DEMATEL model, and the Entropy method, along with the expert evaluation process, to ensure methodological transparency. We believe these revisions improve the clarity and rigor of the study’s design, questions, hypotheses, and methods.

Are the arguments and discussion of findings coherent, balanced and compelling?

Must be improved

Thank you for this valuable comment. In response, we have substantially revised the Discussion section to improve the coherence and persuasiveness of our arguments. The section is now more clearly structured, with findings explicitly linked to both theoretical perspectives and recent empirical studies in sustainability research. We also balanced the discussion by acknowledging potential limitations and outlining implications for both academia and practice. In addition, a dedicated subsection on future research directions has been included to ensure a more comprehensive treatment of the findings. We believe these improvements have enhanced the overall clarity, balance, and impact of the discussion.

For empirical research, are the results clearly presented?

Can be improved

Thank you for this helpful comment. We have revised the Results to present the empirical findings more clearly and transparently. First, manuscript is reorganized and we integrated sensitivity/robustness routines into the analysis software and documented them in Appendix B: k-fold cross-validation for generalizability; architectural sensitivity (layers/neurons/activations); noise-robustness tests via Gaussian perturbations; local sensitivity via Jacobian norms; and stability checks on direct-relation matrices via Frobenius norms. We also report MSE and MAE to summarize predictive error and include compact plots summarizing variability across folds and perturbation levels. We believe these additions make the empirical results clearer, reproducible, and easier to interpret.

Is the article adequately referenced?

Can be improved

Thank you for this important comment. In response, we have substantially expanded and updated the reference list to ensure comprehensive coverage of both classical and recent works in sustainability and multi-criteria decision-making. Several new references from the Sustainability journal and other high-impact outlets have been added to strengthen the theoretical foundation and contextualize our findings within the current literature. In addition, company reports were formatted in line with MDPI standards to ensure consistency and accuracy. We believe these improvements have significantly enhanced the quality, relevance, and adequacy of the referencing throughout the manuscript.

Are the conclusions thoroughly supported by the results presented in the article or referenced in secondary literature?

Can be improved

Thank you for this valuable comment. In response, we have revised the Conclusions section to ensure that the main points are clearly supported by the results and consistent with the discussion. The conclusions are now explicitly linked to the research questions and hypotheses formulated in the study, showing how the findings address and support them. In addition, we incorporated relevant literature to reinforce our interpretations and added implications for practice and future research. We believe these revisions make the conclusions clearer, more coherent, and convincingly grounded in both our results and the broader scholarly context.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1:

The overall content and structure reveal some evident deficiencies. A substantial portion of the manuscript is occupied by extensive data listings and Python code, which, although reflecting the authors’ conscientiousness and openness, distracts considerably from the central research question and undermines the clarity and scholarly value of the paper. The inclusion of such material in the main text does not add academic merit; rather, it dilutes the focus of the study.

Response 1:

We sincerely thank the reviewer for this important observation regarding the balance between methodological transparency and maintaining scholarly focus. In response, we carefully revised the manuscript to ensure that the presentation of data and technical details no longer distracts from the central research question. The following measures were taken:

1. Relocating Extensive Data Listings
• Detailed firm-level performance datasets that were previously embedded within the main text have now been transferred to Appendix A.
• This restructuring allows the main body of the article to highlight only the most relevant findings while still ensuring full transparency for readers who wish to examine the raw data.
2. Relocating Python Code
• The Python implementation of the ANN-based reverse DEMATEL framework has been moved from the main text to Appendix B.
• In the revised manuscript, we provide only a concise conceptual overview of the algorithm in Section 2 (Materials and Methods), while referring readers to the appendix for the complete technical details. This ensures methodological reproducibility without compromising readability.
3. Strengthening Scholarly Focus
• The core narrative of the manuscript has been streamlined to emphasize the research objectives, hypotheses, and contributions rather than lengthy data presentation.
• We restructured the sections to ensure that the main text concentrates on conceptual framing, methodological rationale, and analytical insights, while supplementary details are made available in appendices.
4. Enhancing Clarity and Readability
• To replace long textual descriptions of technical processes, a visual workflow diagram was added to illustrate the role of the ANN within the overall research framework.
• This adjustment improves comprehension for readers unfamiliar with AI-based MCDM techniques and keeps the focus on the study’s theoretical and practical contributions.

Through these revisions, we believe that the manuscript now maintains both academic rigor and readability. The central research question is presented more clearly, the contribution of the study is easier to follow, and the extensive technical and data-related material has been properly relocated to appendices where it can serve as a resource without overwhelming the main discussion.

Comments 2:

Judging from the title and introduction, the manuscript seeks to propose a new methodological approach. However, the development of a new method in the academic context requires a particularly rigorous foundation. The field of Multi-Criteria Decision-Making (MCDM) already hosts a wide variety of established approaches. The authors therefore need to conduct a systematic and comprehensive literature review to accurately delineate the major research gaps that their proposed method is intended to address. Without such groundwork, the novelty and necessity of the study remain insufficiently justified.

Response 2:

We sincerely thank the reviewer for this important observation regarding the need for a stronger conceptual foundation when proposing a novel methodological approach. In response, we substantially expanded and reorganized the Literature Review and Materials and Methods sections to more rigorously justify the contribution of our ANN-based reverse DEMATEL framework. The following revisions were made:

1. Expanded Literature Review
• A comprehensive review of approximately 20 peer-reviewed studies applying MCDM techniques to sustainability contexts was added.
• These include applications of Entropy, DEMATEL, AHP, TOPSIS, PROMETHEE, and hybrid models in areas such as energy planning, green supplier evaluation, transport systems, and corporate sustainability performance.
• Each study is summarized with its scope, methodology, and contribution, highlighting both the strengths and limitations of existing approaches.
2. Identification of Research Gaps
• The revised section explicitly demonstrates that while Entropy provides objective weights, it lacks causal interpretation, and while DEMATEL reveals cause–effect structures, it relies on subjective expert judgment.
• We then show how existing hybrid methods partly address these issues but still depend on subjective panels or are not easily scalable.
• This gap motivates our proposal of an ANN-assisted reverse DEMATEL method that computationally reconstructs causal interdependencies from objective entropy-derived weights.
3. Stronger Link to Research Objectives and Hypotheses
• The extended literature review is now directly connected to the study’s research questions and hypotheses, showing how prior findings informed our focus on the bidirectional interactions between R&D expenditures and sustainability indicators.
• This connection ensures that the novelty and necessity of our study are grounded in established scholarship and clearly positioned within the ongoing development of MCDM methodologies.

Through these revisions, we believe that the manuscript now establishes a far more rigorous conceptual foundation. The expanded literature review not only situates our work within the broader MCDM tradition but also makes explicit the specific methodological gap that our proposed ANN-based reverse DEMATEL framework is designed to address.

Comments 3:

A methodological contribution requires thorough validation to demonstrate its effectiveness, feasibility, and potential priority over existing alternatives. At present, the manuscript does not include convincing case studies, numerical experiments, or illustrative examples that would allow readers to assess the practical value of the proposed approach. Such applications are indispensable for establishing the scholarly contribution of the work.

Response 3:

We thank the reviewer for raising the important point that a methodological contribution requires thorough validation to establish its effectiveness and scholarly value. In response, we significantly strengthened the manuscript by incorporating multiple forms of validation and illustrative applications:

1. Firm-Level Case Applications
• We applied the proposed ANN-based reverse DEMATEL framework to ten firms listed in the Borsa Istanbul Sustainability Index, covering a five-year period and spanning diverse industries.
• For each company, environmental and financial sustainability indicators were analyzed in relation to R&D expenditures, generating direct-relation matrices and cause–effect diagrams.
• These applications serve as real-world case studies that concretely demonstrate how the method can capture systemic interdependencies and provide actionable insights.
2. Numerical Experiments and Comparative Analysis
• The entropy-derived weights were reconstructed into direct-relation matrices, enabling a structured comparison with classical DEMATEL outputs.
• This allowed us to evaluate how well the reverse-engineered model reproduces influence structures consistent with established theory.
• The results highlight meaningful bidirectional relationships—e.g., between R&D and greenhouse gas emissions or profitability—thus validating the model’s interpretive power.
3. Robustness and Sensitivity Tests
• To further assess feasibility and stability, we conducted architectural sensitivity checks (varying layers, neurons, activation functions), noise robustness tests, and k-fold cross-validation.
• Evaluation metrics (MSE, MAE, Jacobian norms, Frobenius norms) confirmed that the model delivers stable and reliable outputs across different configurations and perturbations.
• These results are documented in Appendix B, offering readers transparent access to the technical validation process.

Through these enhancements, the manuscript now provides not only a methodological proposal but also rigorous validation through real corporate case studies, numerical experimentation, and robustness analysis. We believe these additions address the reviewer’s concerns by demonstrating the feasibility, practical utility, and scholarly contribution of the proposed approach.

Comments 4:

The sections devoted to methodology and discussion lack the necessary depth and analytical rigor. These parts should be significantly expanded and articulated with greater precision, so as to strengthen the theoretical underpinnings and practical implications of the study.

Response 4:

We sincerely thank the reviewer for emphasizing the need for greater analytical rigor and precision in the methodology and discussion sections. In response, we have substantially revised and expanded both parts of the manuscript:

1. Methodology Enhancements
• We restructured the Materials and Methods section to include a comprehensive literature-grounded explanation of entropy and DEMATEL applications in sustainability contexts. This now provides a stronger theoretical basis for our approach.
• A dedicated subsection on “Research Questions and Hypotheses” was added, clearly linking our innovation-centered perspective with testable assumptions regarding R&D’s bidirectional role in environmental and financial performance.
• We introduced a “Sample and Data Design” subsection to explicitly describe firm selection criteria, observation periods, inclusion/exclusion rules, and data sources, thereby reinforcing transparency and methodological clarity.
• To increase accessibility, we streamlined the technical description of the ANN-based reverse DEMATEL model while retaining robustness. Supplementary Python code and extended tables were moved to Appendix B, ensuring conciseness in the main text.
2. Discussion Enhancements
• The Discussion section now explicitly connects the results to the hypotheses formulated earlier, clarifying whether findings validate or challenge our assumptions.
• We expanded the section by comparing our findings with prior studies in sustainability MCDM literature (e.g., Bai et al., 2020; Tseng et al., 2019; Govindan et al., 2015), situating our contributions within the broader academic debate.
• The limitations of the work were articulated in a new dedicated paragraph, addressing the scope of inference, methodological assumptions, and directions for future research.
• We emphasized policy, managerial, and sustainability implications, showing how the ANN-assisted DEMATEL framework goes beyond descriptive evaluation to inform strategic decision-making in practice.

Through these revisions, the methodology and discussion sections now integrate stronger theoretical underpinnings, clearer analytical framing, and more robust contextualization within the literature. We believe these improvements address the reviewer’s concerns and substantially strengthen the scholarly contribution of the manuscript.

Comments 5:

The references contain numerous non-English entries, and the citation style is inconsistent and frequently incorrect. The authors are strongly encouraged to ensure that sources are properly formatted and that references to data are accompanied by clear pathways for external verification, such as URLs or other publicly accessible links.

Response 5:

Thank you very much for your detailed and constructive feedback on the citations and references. We fully acknowledge the issues raised and have undertaken a comprehensive revision of the entire References section and in-text citations to ensure full alignment with the journal’s standards. The following steps have been implemented to address each of your observations:

1. Compliance with journal citation standards
   All references and in-text citations have been reformatted according to the journal’s guidelines. This includes consistency in author names, publication years, use of italics for journal titles, inclusion of volume/issue and page numbers, punctuation, and placement of DOIs. The manuscript was carefully checked line by line to ensure that the reference style now fully complies with MDPI’s format.
2. Cross-checking of in-text citations and reference list
   We identified and corrected instances where references appeared in the list but were not cited in the body text (e.g., Harari & Cohen, 2023). Each reference was checked against the text, and unnecessary or uncited items were removed. Conversely, missing references for works cited in the body have been added. As a result, the reference list is now in complete alignment with the manuscript content.
3. Expansion of scholarly bibliography
   To address the concern that the bibliography was too limited, we substantially enriched the list by incorporating additional high-quality, recent studies from journals such as Sustainability, Journal of Cleaner Production, Technological Forecasting and Social Change, and Renewable and Sustainable Energy Reviews. These additions strengthen the theoretical background, expand the scope of the literature review, and situate our study more firmly within current academic debates.
4. Verification and correction of DOIs
   Every single DOI was systematically checked for accuracy. References without DOIs or with incorrect DOIs (e.g., Kumar & Gupta, 2023) were either corrected using official sources (CrossRef, publisher websites) or replaced with verifiable references. Where DOIs were not available (e.g., company reports), we provided stable URLs and proper access dates. This ensures that all references are traceable and verifiable by readers.
5. Correction of incomplete references
   We carefully reviewed the section of the reference list from number 33 onward, where incomplete entries had previously only included author names and years. All of these references have now been updated with complete bibliographic details, including titles, journal names, volumes, issues, page ranges, and DOIs/URLs. This correction ensures that every entry in the reference list is complete and accurate.
6. Standardization of non-article references
   For company reports, financial statements, and sustainability reports used in the study, we reformatted the citations in line with MDPI’s standards. Specifically, the English title is presented first, followed by the original Turkish title in brackets. Publisher/corporate information (e.g., “Ülker: Istanbul, Turkey”) has been added, and each entry includes a stable online link and an access date (e.g., “Available online: … (accessed on 14 April 2025)”). This standardization ensures that all non-academic sources are cited in a professional and consistent manner.
7. Overall improvements in accuracy and coherence
   After these revisions, the reference list is now accurate, comprehensive, and up to date. Each reference cited in the text corresponds to a complete bibliographic entry, and the list adequately reflects the breadth of theoretical and empirical works relevant to our study. In addition, the revisions improved the balance between classical foundational works and recent contributions, thereby enhancing the academic rigor of the manuscript.

In summary, we have carried out a thorough revision of the references to address all concerns raised. We are confident that the revised manuscript now meets the journal’s requirements for accurate, comprehensive, and properly formatted citations and references, while also providing a stronger and more credible scholarly foundation for our research.

Comments 6:

Perhaps the most serious issue lies in the overall organization of the paper, which does not conform to the established conventions of academic research articles. The authors should carefully examine high-quality methodological research papers in the field and restructure their manuscript accordingly, following a more conventional academic paradigm.

Response 6:

We sincerely thank the reviewer for pointing out the need to align the manuscript more closely with established academic conventions. In response, we carefully restructured the paper to follow a more conventional and rigorous academic paradigm, drawing on best practices observed in high-quality methodological research articles in the MCDM field.

1. Improved Section Organization
• We introduced a dedicated Literature Review section at the beginning of Materials and Methods, systematically discussing prior applications of entropy, DEMATEL, and hybrid MCDM models in sustainability contexts, and positioning our work within these streams.
• A new paragraph for “Research Questions and Hypotheses” was added to Section 2, explicitly formulating our guiding research questions and hypotheses.
• A “Sample and Data Design” paragraph was incorporated, detailing sectoral coverage, observation period, selection criteria, and data sources, thereby clarifying the scope and design logic of the study.
2. Refined Methodology and Results Separation
• The Materials and Methods section was revised for conciseness by moving detailed tables and technical code descriptions to the Appendices, while retaining only the essential methodological narrative in the main text.
• We created a standalone Conclusions section (Section 3) to clearly separate empirical findings from the methodological description.
3. Enhanced Discussion and Conclusions
• The Discussion section was expanded and directly linked to the literature review, with explicit validation (or rejection) of the proposed hypotheses and comparison with related studies in other markets.
• A dedicated Conclusions section was introduced, synthesizing key findings, contributions, and implications.
• Finally, a Limitations of the Work paragraph was added to acknowledge the scope and boundaries of our study while suggesting directions for future research.

Through these structural and organizational revisions, we believe the manuscript now conforms more closely to the expectations of methodological research articles, offering improved clarity, rigor, and readability while preserving the novelty of our ANN-based reverse DEMATEL approach.

Comments 7:

In summary, while the authors’ effort and transparency are appreciable, the manuscript requires substantial revisions in terms of structure, rigor, and validation to achieve the level of quality expected for scholarly publication.

Response 7:

We sincerely thank the reviewer for their constructive overall assessment of the manuscript. We carefully considered the concerns raised and undertook substantial revisions to strengthen the structure, rigor, and validation of the study, ensuring it meets the standards of scholarly publication.

1. Structural Improvements
• We restructured the manuscript following conventional academic norms by clearly distinguishing the Introduction, Literature Review, Materials and Methods, Conclusions and Discussion, sections.
• A dedicated Literature Review was added, systematically mapping prior applications of entropy, DEMATEL, and hybrid MCDM approaches in sustainability contexts. This establishes a stronger conceptual foundation and clarifies the research gaps our study addresses.
• A Research Questions and Hypotheses subsection was introduced, explicitly linking the study’s objectives to testable propositions.
• A Sample and Data Design subsection was added to detail firm selection, period, and inclusion criteria.
2. Enhanced Rigor
• The Materials and Methods section was revised for clarity and accessibility: several technical tables and detailed code descriptions were moved to the Appendices, while the main text was streamlined with clearer narratives and methodological workflow figures.
• We highlighted the theoretical integration of entropy (objective weighting) and DEMATEL (causal mapping) and emphasized how the ANN-based reverse DEMATEL framework bridges their limitations.
• The Discussion was significantly expanded and now directly connects results to the literature, validating or rejecting our hypotheses, and comparing findings with studies in other markets.
3. Validation and Practical Value
• To demonstrate practical applicability, we applied the proposed framework to a five-year dataset of ten firms listed on the Borsa Istanbul Sustainability Index, covering diverse sectors.
• We incorporated robustness and sensitivity analyses (Appendix B), testing architectural sensitivity, cross-validation, noise robustness, and Jacobian-based local sensitivity to ensure methodological soundness.
• We added a Limitations of the Work subsection, acknowledging the scope of the analysis and proposing avenues for extending the method in future research.

Through these comprehensive revisions, the manuscript now provides (i) a rigorous conceptual basis, (ii) a transparent methodological contribution, and (iii) empirically grounded validation, ensuring clarity and scholarly value. We believe the revised version adequately addresses the reviewer’s concerns and substantially enhances the overall quality of the paper.

4. Response to Comments on the Quality of English Language

Point 1:

The English is fine and does not require any improvement.

Response 1:

Thank you for your positive remark regarding the language quality of our manuscript. Following the overall revisions made throughout the paper, we have also made minor language refinements to ensure consistency and improve clarity. These edits were carried out with the support of a native English speaker to enhance fluency and readability. We believe these adjustments contribute positively to the overall presentation of the manuscript.

5. Additional clarifications

We would like to sincerely thank the reviewer and the editorial team for their insightful and constructive feedback. The comments provided have significantly contributed to the development and refinement of our manuscript. We have addressed all suggestions to the best of our ability, and we genuinely hope that the revised version meets your expectations.

Response to Reviewer 3 Comments

1. Summary

We would like to express our sincere gratitude for the time and effort you devoted to reviewing our manuscript. Your constructive comments and insightful recommendations have been invaluable in strengthening both the quality and presentation of our work. In response, we have undertaken several key revisions. Specifically, we elaborated the Introduction to provide clearer context and research motivation, refined the Methodology section with more transparent descriptions of data collection and analysis procedures, and expanded the Results and Discussion with additional comparisons to recent studies in the field. We also improved the readability of the manuscript by restructuring some paragraphs, enriching the figures and tables for better clarity, and updating the reference list to include recent and relevant contributions. We trust that these revisions have enhanced the rigor, coherence, and contribution of the study.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Is the content succinctly described and contextualized with respect to previous and present theoretical background and empirical research (if applicable) on the topic?

Must be improved

Thank you for your valuable comment. In response, we have revised Sections 1 and 2 to strengthen the theoretical background and to provide a clearer link between prior research and the present study. We believe these revisions improve the contextualization and address your concern.

Are the research design, questions, hypotheses and methods clearly stated?

Must be improved

Thank you for this important comment. In response, we have revised manuscript to more clearly present the research design and methodological framework. We explicitly stated the research questions and formulated hypotheses to guide the analysis, thereby strengthening the link between theory and empirical investigation. Furthermore, we elaborated on the use of DEMATEL, the ANN-based reverse DEMATEL model, and the Entropy method, along with the expert evaluation process, to ensure methodological transparency. We believe these revisions improve the clarity and rigor of the study’s design, questions, hypotheses, and methods.

Are the arguments and discussion of findings coherent, balanced and compelling?

Must be improved

Thank you for this valuable comment. In response, we have substantially revised the Discussion section to improve the coherence and persuasiveness of our arguments. The section is now more clearly structured, with findings explicitly linked to both theoretical perspectives and recent empirical studies in sustainability research. We also balanced the discussion by acknowledging potential limitations and outlining implications for both academia and practice. In addition, a dedicated subsection on future research directions has been included to ensure a more comprehensive treatment of the findings. We believe these improvements have enhanced the overall clarity, balance, and impact of the discussion.

For empirical research, are the results clearly presented?

Can be improved

Thank you for this helpful comment. We have revised the Results to present the empirical findings more clearly and transparently. First, manuscript is reorganized and we integrated sensitivity/robustness routines into the analysis software and documented them in Appendix B: k-fold cross-validation for generalizability; architectural sensitivity (layers/neurons/activations); noise-robustness tests via Gaussian perturbations; local sensitivity via Jacobian norms; and stability checks on direct-relation matrices via Frobenius norms. We also report MSE and MAE to summarize predictive error and include compact plots summarizing variability across folds and perturbation levels. We believe these additions make the empirical results clearer, reproducible, and easier to interpret.

Is the article adequately referenced?

Must be improved

Thank you for this important comment. In response, we have substantially expanded and updated the reference list to ensure comprehensive coverage of both classical and recent works in sustainability and multi-criteria decision-making. Several new references from the Sustainability journal and other high-impact outlets have been added to strengthen the theoretical foundation and contextualize our findings within the current literature. In addition, company reports were formatted in line with MDPI standards to ensure consistency and accuracy. We believe these improvements have significantly enhanced the quality, relevance, and adequacy of the referencing throughout the manuscript.

Are the conclusions thoroughly supported by the results presented in the article or referenced in secondary literature?

Must be improved

Thank you for this valuable comment. In response, we have revised the Conclusions section to ensure that the main points are clearly supported by the results and consistent with the discussion. The conclusions are now explicitly linked to the research questions and hypotheses formulated in the study, showing how the findings address and support them. In addition, we incorporated relevant literature to reinforce our interpretations and added implications for practice and future research. We believe these revisions make the conclusions clearer, more coherent, and convincingly grounded in both our results and the broader scholarly context.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1:

Title

• The title is confusing.

Response 1:

We would like to begin by sincerely thanking the esteemed reviewer for their valuable and constructive comments regarding the complexity of our original title. Your feedback has been instrumental in helping us refine and improve the clarity of our manuscript.

First Title:
‘’Investigation of Corporate Sustainability Performance Data and Developing an Innovation-Oriented Novel Analysis Method with Multi-Criteria Decision Making Approach’’

Revised Title:
‘’Innovation-Oriented Novel Multi-Criteria Decision Analysis of Corporate Sustainability: Evidence from the Türkiye Sustainability Index’’

The original title, while comprehensive, contained multiple consecutive noun phrases and a relatively high character count. This structure made it appear more like a thesis statement than a concise article title.

The revised version has fewer compound noun phrases, uses fewer words overall, and is significantly shorter in terms of character count. This results in a smoother reading flow and enhanced accessibility for an international audience.

The revised title adopts the widely accepted “X of Y: Evidence from Z” format, which is frequently used in high-quality journals. This makes it not only more recognizable but also easier to cite and recall.

We sincerely thank the reviewer once again for pointing out the issue regarding the complexity of the original title. In response, we have revised it to:

“Innovation-Oriented Novel Multi-Criteria Decision Analysis of Corporate Sustainability: Evidence from the Türkiye Sustainability Index”

This revised version is shorter, clearer, and more fluent, while still reflecting the methodological novelty and empirical contribution of our study. We believe this adjustment directly addresses the reviewer’s concern and enhances the readability and impact of the manuscript.

Comments 2:

Abstract
• A brief contextualization or justification of the relevance of the topic is missing.
• The author does not formulate the objectives of the work, but rather describes what is done. It is suggested to briefly describe the objective(s) of the work.
• The authors state: “Using the entropy method from the Multi-Criteria Decision Making (MCDM) framework, objective weights of sustainability-related criteria were calculated.” According to this statement, the work would consist solely of performing a calculation.
• Acronyms are used without indicating what they stand for, DEMATEL. This is not indicated anywhere in the document.

Response 2:

We would like to sincerely thank the esteemed reviewer for the constructive feedback regarding the abstract section. In line with the comments, we have undertaken a number of improvements to enhance both clarity and academic rigor:

1. Contextualization of the Topic
   In the revised abstract, we added a clear justification for the relevance of our study by highlighting the increasing importance of aligning innovation with corporate sustainability strategies. This ensures that the abstract not only describes what was done but also explains why the research matters in a broader societal and managerial context.
2. Explicit Statement of Objectives
   The original abstract primarily described the methodological steps. Following the reviewer’s suggestion, we have now explicitly articulated the main objective of the study, namely: to develop and test a novel, data-driven analytical framework that reduces reliance on subjective expert judgments while providing actionable insights for sustainability-oriented decision-making.
3. Clarification of Methodological Contribution
   To address the concern that the abstract might suggest the work consisted solely of a calculation using the entropy method, we clarified that entropy weighting represents the first stage of the analysis. We emphasized that the novel contribution lies in the development of an artificial neural network–based reverse DEMATEL model, which enables the reconstruction of inter-criteria relationships and the exploration of systemic sustainability dynamics.
4. Innovation and R&D Emphasis
   Given the central role of innovation in our research, we strengthened the description of how R&D expenditures function both as drivers and as outcomes within sustainability interactions. This was highlighted as a unique contribution of the proposed approach.
5. Terminological Consistency
   In response to the reviewer’s observations, we also ensured terminological clarity. We added the full form of DEMATEL (Decision-Making Trial and Evaluation Laboratory) in the Abbreviations section of the manuscript, to ensure international readability and precision.

Summary of Changes
The original abstract largely focused on describing the methods applied, while the revised version now (i) contextualizes the importance of the research, (ii) explicitly states the objectives, (iii) clarifies the methodological novelty beyond entropy weighting, and (iv) highlights the central role of innovation and R&D. These revisions make the abstract more concise, academically precise, and aligned with the reviewer’s recommendations.

Comments 3:

Introduction

In this section, the authors: 1) describe the topic of the work; 2) explain the critical gap that exists in these matters; 3) indicate that the work addresses that gap; 4) provide a very general description of the methodology used; 5) include a reflection on the contribution of the work in the final paragraph. However, this section should also include:
• A description of the specific objectives of the work.
• A better description of the methodology.
• A brief description of the main results.
• A final paragraph summarizing the content of the other sections of the document.

Response 3:

We would like to express our sincere gratitude to the reviewer for their constructive and valuable feedback regarding the Introduction section. In response, we carefully revised this part of the manuscript to incorporate the requested improvements. Below, we detail the specific changes that were made:

1. Clear Statement of Objectives

·         Reviewer’s Comment: The introduction should include a description of the specific objectives of the work.

·         Our Revision: A dedicated paragraph was added explicitly stating the main objective of the study: to design and apply a novel reverse DEMATEL approach, supported by artificial neural network–based software, that reconstructs inter-criteria influence matrices from Entropy-derived weights. This clarified the purpose of the work and its intended contribution to sustainability research.

2. Better Description of the Methodology

·         Reviewer’s Comment: A better description of the methodology should be included.

·         Our Revision: The revised introduction now provides a concise description of the methodological framework, summarizing the following steps:

1.       Five-year sustainability performance data were collected from ten companies listed in the Borsa Istanbul Sustainability Index.

2.       Entropy was applied to calculate objective criterion weights.

3.       Reverse DEMATEL was implemented using custom ANN-based software to reconstruct direct influence matrices.

4.       R&D expenditures were integrated as a central bridging criterion, enabling systemic analysis of sustainability interactions.
This addition strengthens the reader’s understanding of the methodological innovation presented in the study.

3. Brief Description of Main Results

·         Reviewer’s Comment: The introduction should include a brief description of the main results.

·         Our Revision: A summary of the main findings was incorporated, highlighting that:

o    Environmental indicators (e.g., greenhouse gas emissions, energy consumption) strongly influence R&D expenditures.

o    Financial indicators (e.g., profitability measures) are more significantly shaped by R&D rather than driving it.

o    This asymmetry underscores the role of R&D as a central driver of both financial and environmental performance.

o    Firm-specific strategic insights were also identified, such as the prioritization of emission-focused innovation in high-pollution industries.

4. Summary of the Paper Structure

·         Reviewer’s Comment: The introduction should end with a paragraph summarizing the structure of the manuscript.

·         Our Revision: A new paragraph was added that clearly outlines the content of each section: methodology in Section 2, results in Section 3, discussion in Section 4, and conclusions in Section 5. This provides readers with a roadmap of the paper.

5. Preservation of Referenced Literature

·         All previously cited sources and references were preserved as requested, ensuring the revised introduction maintains consistency with the original scholarly grounding.

Through these revisions, the Introduction section now not only establishes the context and critical research gap but also explicitly states the objectives, methodology, main findings, and manuscript structure. We believe these improvements address the reviewer’s concerns and enhance the clarity, readability, and academic rigor of the paper.

Comments 4:

Materials and Methods
• The document does not include a literature review section.
• Very few academic works are cited. The article lacks a conceptual basis. The bibliographic background is limited.
• No hypotheses or relationships between variables are formulated.
• The sample design is not explained.
• The document consists of applying a calculation method. Most of the document is devoted to presenting data from the selected companies, presenting the results of the calculations, and describing those results.

Response 4:

We would like to express our sincere gratitude to the reviewer for their constructive and insightful feedback regarding the Materials and Methods section. In response, we carefully revised this part of the manuscript to address the requested improvements and strengthen the conceptual and methodological rigor. Below, we outline the specific changes and additions that were made.

1- 

• We have developed a comprehensive literature review section that synthesizes academic studies employing Multi-Criteria Decision-Making (MCDM) approaches in sustainability contexts.
• This section explains how entropy, DEMATEL, TOPSIS, ANP, PROMETHEE, and hybrid methods have been applied in prior research.
• It highlights two recurring patterns (objective weighting vs. subjective causality mapping) and positions our contribution—the ANN-assisted reverse DEMATEL—as a methodological bridge.

2- 

• The revised manuscript now integrates a substantially expanded set of references from high-impact journals (e.g., Journal of Cleaner Production, Expert Systems with Applications, Renewable and Sustainable Energy Reviews, Sustainability).
• We explicitly discuss the conceptual basis of our work by linking corporate sustainability, innovation (R&D), and MCDM methods.
• The bibliographic background now provides both theoretical grounding and practical context for why a new methodology is needed.

3- 

• We have added a dedicated subsection which includes “Research Questions and Hypotheses.” after literature review.
• This section formulates explicit research questions (RQ1, RQ2) and hypotheses (H1–H5) regarding the bidirectional interactions between R&D and sustainability criteria.
• These hypotheses clarify expected causal directions (e.g., R&D reducing GHG emissions; profitability enabling R&D) and guide the interpretation of results.

4-

• A new explanation which includes “Sample and Data Design” has been introduced in the end of Materials and Methods section.
• This clarifies the inclusion criteria: (i) firms continuously listed in the BIST Sustainability Index (2018–2022); (ii) full disclosure of environmental and financial indicators; (iii) availability of R&D data.
• We explain the sectoral diversity of the ten firms, the data sources (sustainability, integrated, financial reports), and the conservative handling of missing values.
• This ensures transparency and rigor in how the empirical sample was constructed.

5- 

• To counter this impression, we added the Analytical Workflow explanation at the end of Materials and Methods section.
• The revised manuscript clarifies that our contribution goes beyond calculation:
• Entropy weighting provides objective importance scores.
• ANN-assisted reverse DEMATEL reconstructs directional interdependencies without expert bias.
• Robustness checks (architectural sensitivity, cross-validation, noise robustness, Jacobian analysis) validate the reliability of the model.
• A conceptual flowchart is planned to visually demonstrate the stepwise pipeline (data → entropy → reverse DEMATEL → influence digraphs → robustness diagnostics).
• Thus, the manuscript now emphasizes methodological innovation and systemic modeling, rather than just numerical reporting.

Comments 5:

Results and Discussion
• The content of this section is not a presentation of the results or a discussion of the results.
• The content of this section should be included in the conclusions section.
Conclusions
• The authors do not include a conclusions section. This is unacceptable in a scientific article. The authors have material to write this section based on the content of the Results and Discussion section.
• The limitations of the work are missing.

Response 5:

We sincerely thank the reviewer for their constructive feedback on the Conclusions and Discussion sections. In response, we revised these parts to clearly synthesize the main findings, emphasize their theoretical and practical implications, and add a dedicated subsection on the study’s limitations. These improvements ensure that the manuscript now presents a balanced conclusion and a more focused, critical discussion in line with the reviewer’s guidance.

1- In the revised manuscript, we restructured the Discussion section to ensure it is focused on analyzing and interpreting the results, rather than simply presenting them. Explanations that were more descriptive in nature have now been relocated to the Results section, while the Discussion emphasizes interpretation, strategic implications, and theoretical contributions. This revision ensures a clearer separation between results presentation and critical discussion.

2- We revised the manuscript so that descriptive summaries previously included in the Discussion have been moved into the newly structured Conclusions section. This helps avoid redundancy and provides a more logical flow. The Conclusions section now includes a clear synthesis of findings, their implications, and the broader contributions of the study.

3-  In the revised manuscript, we introduced a dedicated Conclusions section. This section now synthesizes the key findings derived from the analysis, outlines policy and managerial implications, and highlights how R&D functions as a bridging factor between environmental and financial sustainability. We also summarize the added value of our novel ANN-based reverse DEMATEL approach and its contributions to both methodological innovation and practical strategy design. The section has been developed directly from the material of the Results and Discussion, as suggested.

4- A subsection titled Limitations of the Work has now been added to the end of the Discussion section. This part explicitly addresses the boundaries of our research design, including the scope of firms analyzed, the ANN-based reverse DEMATEL framework’s simulated nature, and the potential to extend the methodology to social sustainability dimensions. We framed these points in a constructive manner, showing how they open opportunities for future research while maintaining the robustness of our current findings.

Comments 6:

Citations and References
• The way references are cited does not comply with the journal's standards.
• Several references were not found in the body of the document. For example: Harari, Y., & Cohen, D. (2023). Decision support systems for sustainability transitions. Systems Research and Behavioral Science, 40(2), 187–201.
• The scholarly bibliography is very limited.
• Some articles do not have a DOI. For example, the following article does not have a DOI. I couldn't find it by title or journal: Kumar, R., & Gupta, V. (2023). Structural influence of innovation criteria via hybrid MCDM. Journal of Business Research, 168, 487–500.
• From reference 33 onward, only the authors' names and the year of publication are indicated. For this reason, I was unable to analyze a significant portion of the document

Response 6:

Thank you very much for your detailed and constructive feedback on the citations and references. We fully acknowledge the issues raised and have undertaken a comprehensive revision of the entire References section and in-text citations to ensure full alignment with the journal’s standards. The following steps have been implemented to address each of your observations:

1. Compliance with journal citation standards
   All references and in-text citations have been reformatted according to the journal’s guidelines. This includes consistency in author names, publication years, use of italics for journal titles, inclusion of volume/issue and page numbers, punctuation, and placement of DOIs. The manuscript was carefully checked line by line to ensure that the reference style now fully complies with MDPI’s format.
2. Cross-checking of in-text citations and reference list
   We identified and corrected instances where references appeared in the list but were not cited in the body text (e.g., Harari & Cohen, 2023). Each reference was checked against the text, and unnecessary or uncited items were removed. Conversely, missing references for works cited in the body have been added. As a result, the reference list is now in complete alignment with the manuscript content.
3. Expansion of scholarly bibliography
   To address the concern that the bibliography was too limited, we substantially enriched the list by incorporating additional high-quality, recent studies from journals such as Sustainability, Journal of Cleaner Production, Technological Forecasting and Social Change, and Renewable and Sustainable Energy Reviews. These additions strengthen the theoretical background, expand the scope of the literature review, and situate our study more firmly within current academic debates.
4. Verification and correction of DOIs
   Every single DOI was systematically checked for accuracy. References without DOIs or with incorrect DOIs (e.g., Kumar & Gupta, 2023) were either corrected using official sources (CrossRef, publisher websites) or replaced with verifiable references. Where DOIs were not available (e.g., company reports), we provided stable URLs and proper access dates. This ensures that all references are traceable and verifiable by readers.
5. Correction of incomplete references
   We carefully reviewed the section of the reference list from number 33 onward, where incomplete entries had previously only included author names and years. All of these references have now been updated with complete bibliographic details, including titles, journal names, volumes, issues, page ranges, and DOIs/URLs. This correction ensures that every entry in the reference list is complete and accurate.
6. Standardization of non-article references
   For company reports, financial statements, and sustainability reports used in the study, we reformatted the citations in line with MDPI’s standards. Specifically, the English title is presented first, followed by the original Turkish title in brackets. Publisher/corporate information (e.g., “Ülker: Istanbul, Turkey”) has been added, and each entry includes a stable online link and an access date (e.g., “Available online: … (accessed on 14 April 2025)”). This standardization ensures that all non-academic sources are cited in a professional and consistent manner.
7. Overall improvements in accuracy and coherence
   After these revisions, the reference list is now accurate, comprehensive, and up to date. Each reference cited in the text corresponds to a complete bibliographic entry, and the list adequately reflects the breadth of theoretical and empirical works relevant to our study. In addition, the revisions improved the balance between classical foundational works and recent contributions, thereby enhancing the academic rigor of the manuscript.

In summary, we have carried out a thorough revision of the references to address all concerns raised. We are confident that the revised manuscript now meets the journal’s requirements for accurate, comprehensive, and properly formatted citations and references, while also providing a stronger and more credible scholarly foundation for our research.

4. Response to Comments on the Quality of English Language

Point 1:

The English is fine and does not require any improvement.

Response 1:

Thank you for your positive remark regarding the language quality of our manuscript. Following the overall revisions made throughout the paper, we have also made minor language refinements to ensure consistency and improve clarity. These edits were carried out with the support of a native English speaker to enhance fluency and readability. We believe these adjustments contribute positively to the overall presentation of the manuscript.

5. Additional clarifications

We would like to sincerely thank the reviewer and the editorial team for their insightful and constructive feedback. The comments provided have significantly contributed to the development and refinement of our manuscript. We have addressed all suggestions to the best of our ability, and we genuinely hope that the revised version meets your expectations.

Response to Reviewer 4 Comments

1. Summary

We would like to express our sincere gratitude for the time and effort you devoted to reviewing our manuscript. Your constructive comments and insightful recommendations have been invaluable in strengthening both the quality and presentation of our work. In response, we have undertaken several key revisions. Specifically, we elaborated the Introduction to provide clearer context and research motivation, refined the Methodology section with more transparent descriptions of data collection and analysis procedures, and expanded the Results and Discussion with additional comparisons to recent studies in the field. We also improved the readability of the manuscript by restructuring some paragraphs, enriching the figures and tables for better clarity, and updating the reference list to include recent and relevant contributions. We trust that these revisions have enhanced the rigor, coherence, and contribution of the study.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Is the content succinctly described and contextualized with respect to previous and present theoretical background and empirical research (if applicable) on the topic?

Must be improved

Thank you for your valuable comment. In response, we have revised Sections 1 and 2 to strengthen the theoretical background and to provide a clearer link between prior research and the present study. We believe these revisions improve the contextualization and address your concern.

Are the research design, questions, hypotheses and methods clearly stated?

Must be improved

Thank you for this important comment. In response, we have revised manuscript to more clearly present the research design and methodological framework. We explicitly stated the research questions and formulated hypotheses to guide the analysis, thereby strengthening the link between theory and empirical investigation. Furthermore, we elaborated on the use of DEMATEL, the ANN-based reverse DEMATEL model, and the Entropy method, along with the expert evaluation process, to ensure methodological transparency. We believe these revisions improve the clarity and rigor of the study’s design, questions, hypotheses, and methods.

Are the arguments and discussion of findings coherent, balanced and compelling?

Can be improved

Thank you for this valuable comment. In response, we have substantially revised the Discussion section to improve the coherence and persuasiveness of our arguments. The section is now more clearly structured, with findings explicitly linked to both theoretical perspectives and recent empirical studies in sustainability research. We also balanced the discussion by acknowledging potential limitations and outlining implications for both academia and practice. In addition, a dedicated subsection on future research directions has been included to ensure a more comprehensive treatment of the findings. We believe these improvements have enhanced the overall clarity, balance, and impact of the discussion.

For empirical research, are the results clearly presented?

Yes

Thank you for this helpful and meaningful comment. We also have revised the Results to present the empirical findings more clearly and transparently. First, manuscript is reorganized and we integrated sensitivity/robustness routines into the analysis software and documented them in Appendix B: k-fold cross-validation for generalizability; architectural sensitivity (layers/neurons/activations); noise-robustness tests via Gaussian perturbations; local sensitivity via Jacobian norms; and stability checks on direct-relation matrices via Frobenius norms. We also report MSE and MAE to summarize predictive error and include compact plots summarizing variability across folds and perturbation levels. We believe these additions make the empirical results clearer, reproducible, and easier to interpret.

Is the article adequately referenced?

Must be improved

Thank you for this important comment. In response, we have substantially expanded and updated the reference list to ensure comprehensive coverage of both classical and recent works in sustainability and multi-criteria decision-making. Several new references from the Sustainability journal and other high-impact outlets have been added to strengthen the theoretical foundation and contextualize our findings within the current literature. In addition, company reports were formatted in line with MDPI standards to ensure consistency and accuracy. We believe these improvements have significantly enhanced the quality, relevance, and adequacy of the referencing throughout the manuscript.

Are the conclusions thoroughly supported by the results presented in the article or referenced in secondary literature?

Must be improved

Thank you for this valuable comment. In response, we have revised the Conclusions section to ensure that the main points are clearly supported by the results and consistent with the discussion. The conclusions are now explicitly linked to the research questions and hypotheses formulated in the study, showing how the findings address and support them. In addition, we incorporated relevant literature to reinforce our interpretations and added implications for practice and future research. We believe these revisions make the conclusions clearer, more coherent, and convincingly grounded in both our results and the broader scholarly context.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1:

The literature review is limited. It should be expanded to establish a stronger connection with the research objectives and to support the formulation of hypotheses (in Section 2).

Response 1:

We thank the reviewer for highlighting the need to expand the Literature Review in order to better connect with the research objectives and to support the formulation of hypotheses. In response, we substantially revised Section 2 to provide a more comprehensive conceptual foundation. Specifically:

• We integrated approximately twenty additional academic studies covering diverse applications of MCDM in sustainability contexts (e.g., green supplier evaluation, renewable energy planning, circular economy, eco-innovation strategies). These references establish a broader bibliographic background and situate our work within the global research landscape.
• The revised section now explicitly highlights the complementary roles of entropy (for objective weighting) and DEMATEL (for causal mapping), drawing on prior applications in sustainability to justify our methodological approach.
• Building on this expanded literature base, we formulated a set of targeted Research Questions and Hypotheses (H1–H5) that focus on the bidirectional interactions between R&D expenditures and sustainability indicators. These hypotheses link directly to both the theoretical insights in the literature and the empirical objectives of our study.
• The expanded review and hypotheses together clarify the novelty of our ANN-based reverse DEMATEL framework, while also ensuring that our research objectives are firmly grounded in established scholarship.

We believe these revisions address the reviewer’s concern by strengthening the conceptual rigor, aligning the literature review more clearly with the objectives, and providing a solid foundation for hypothesis formulation.

Comments 2:

The discussion should be better connected with the literature review by justifying the validation (or rejection) of the hypotheses and by comparing results with similar studies conducted in other markets.

Response 2:

We thank the reviewer for pointing out the need to link the Discussion more clearly with the literature review and the validation of hypotheses. Accordingly, we revised the Discussion to explicitly state which hypotheses (H1–H5 in section 2) are supported by our results and to compare our findings with prior studies (e.g., Bai et al., 2020; Govindan et al., 2015; Büyüközkan & Çifçi, 2012). This addition demonstrates that the reciprocal effects observed between R&D and environmental/financial indicators are consistent with established findings in other markets. These revisions ensure that the Discussion is firmly anchored in the literature and that our contributions are positioned within the broader international research context.

Specifically:

• Hypothesis1 was supported, as higher GHG pressures were associated with stronger R&D intensity, and R&D showed mitigating effects on GHG emissions.
• Hypothesis2 was partially supported: energy and water use drove R&D allocation, while R&D in turn reduced intensities, though effects varied across firms.
• Hypothesis3 was validated, since waste and recycled-waste indicators were positively linked with R&D initiatives in circular processes.
• Hypothesis4 was supported: profitability (operating profit, net profit) enabled R&D, while R&D reinforced profitability through efficiency/productivity gains.
• Hypothesis5 was also supported, as structural items like assets and equity showed weaker short-run impacts on R&D, while R&D expenditures contributed to long-term asset strengthening.

Comments 3:

Methodology

• The Materials and Methods section is excessively long. The description of the reverse DEMATEL with neural networks is highly technical and dense, which may reduce accessibility for readers less familiar with AI-based MCDM approaches. Clearer visualizations of the ANN’s role would improve comprehension.
• Several tables could be moved to the appendix to make the main text more concise and reader-friendly.
• Results currently presented within this section should be separated into a dedicated Section 3 -Results

Response 3:

We sincerely thank the reviewer for their constructive feedback regarding the Materials and Methods section. In response, we implemented several targeted revisions to improve clarity, accessibility, and overall structure:

1. Reducing Length and Technical Density
• The original description of the ANN-based reverse DEMATEL framework was highly technical. We carefully revised this section to streamline explanations, reduce unnecessary technical jargon, and emphasize conceptual clarity.
• To improve accessibility for readers less familiar with AI-based MCDM approaches, we restructured the explanation into shorter subsections with clearer step-by-step logic.
2. Improved Visualization of the ANN Role
• We introduced a workflow diagram illustrating the analytical pipeline: from entropy-based weighting, to ANN-assisted reconstruction of direct-relation matrices, to reverse DEMATEL outputs.
• This figure visually clarifies the role of the ANN within the overall framework, making it easier for readers to understand the contribution of this methodological innovation without needing to engage with dense technical detail.
3. Relocating Tables to Appendix
• To make the manuscript more concise and reader-friendly, several lengthy tables that previously appeared in the Materials and Methods section (e.g., detailed entropy weight tables and sensitivity analysis outputs) have been moved to Appendix A and Appendix B.
• In the main text, we retained only summary tables and key results, while providing cross-references to the appendix for readers interested in the complete dataset.
4. Separating Results from Methods
• In the initial submission, some results were embedded within the Materials and Methods section. Following the reviewer’s advice, we reorganized the manuscript to ensure a clearer structure:
• Section 2 (Materials and Methods) now focuses exclusively on data sources, methodological steps, and the ANN-based framework.
• Section 3 (Conclusions) was created as a dedicated section where all outcomes of the entropy weighting, DEMATEL method influence maps, and ANN sensitivity analyses are systematically presented and interpreted.

Through these revisions, we believe the Materials and Methods section has become significantly more concise, accessible, and structured, while the relocation of results and supporting tables ensures smoother readability for the audience.

Comments 4:

References

• The referencing style does not fully conform to the journal’s guidelines and should be corrected accordingly. Many of them are incomplete.

Response 4:

Thank you very much for your constructive feedback on the citations and references. We fully acknowledge the issues raised and have undertaken a comprehensive revision of the entire References section and in-text citations to ensure full alignment with the journal’s standards. The following steps have been implemented to address each of your observations:

1. Compliance with journal citation standards
   All references and in-text citations have been reformatted according to the journal’s guidelines. This includes consistency in author names, publication years, use of italics for journal titles, inclusion of volume/issue and page numbers, punctuation, and placement of DOIs. The manuscript was carefully checked line by line to ensure that the reference style now fully complies with MDPI’s format.
2. Cross-checking of in-text citations and reference list
   We identified and corrected instances where references appeared in the list but were not cited in the body text (e.g., Harari & Cohen, 2023). Each reference was checked against the text, and unnecessary or uncited items were removed. Conversely, missing references for works cited in the body have been added. As a result, the reference list is now in complete alignment with the manuscript content.
3. Expansion of scholarly bibliography
   To address the concern that the bibliography was too limited, we substantially enriched the list by incorporating additional high-quality, recent studies from journals such as Sustainability, Journal of Cleaner Production, Technological Forecasting and Social Change, and Renewable and Sustainable Energy Reviews. These additions strengthen the theoretical background, expand the scope of the literature review, and situate our study more firmly within current academic debates.
4. Verification and correction of DOIs
   Every single DOI was systematically checked for accuracy. References without DOIs or with incorrect DOIs (e.g., Kumar & Gupta, 2023) were either corrected using official sources (CrossRef, publisher websites) or replaced with verifiable references. Where DOIs were not available (e.g., company reports), we provided stable URLs and proper access dates. This ensures that all references are traceable and verifiable by readers.
5. Correction of incomplete references
   We carefully reviewed the section of the reference list from number 33 onward, where incomplete entries had previously only included author names and years. All of these references have now been updated with complete bibliographic details, including titles, journal names, volumes, issues, page ranges, and DOIs/URLs. This correction ensures that every entry in the reference list is complete and accurate.
6. Standardization of non-article references
   For company reports, financial statements, and sustainability reports used in the study, we reformatted the citations in line with MDPI’s standards. Specifically, the English title is presented first, followed by the original Turkish title in brackets. Publisher/corporate information (e.g., “Ülker: Istanbul, Turkey”) has been added, and each entry includes a stable online link and an access date (e.g., “Available online: … (accessed on 14 April 2025)”). This standardization ensures that all non-academic sources are cited in a professional and consistent manner.
7. Overall improvements in accuracy and coherence
   After these revisions, the reference list is now accurate, comprehensive, and up to date. Each reference cited in the text corresponds to a complete bibliographic entry, and the list adequately reflects the breadth of theoretical and empirical works relevant to our study. In addition, the revisions improved the balance between classical foundational works and recent contributions, thereby enhancing the academic rigor of the manuscript.

In summary, we have carried out a thorough revision of the references to address all concerns raised. We are confident that the revised manuscript now meets the journal’s requirements for accurate, comprehensive, and properly formatted citations and references, while also providing a stronger and more credible scholarly foundation for our research.

4. Response to Comments on the Quality of English Language

Point 1:

The English could be improved to more clearly express the research.

Response 1:

Thank you for your positive remark regarding the language quality of our manuscript. Following the overall revisions made throughout the paper, we have also made minor language refinements to ensure consistency and improve clarity. These edits were carried out with the support of a native English speaker to enhance fluency and readability. We believe these adjustments contribute positively to the overall presentation of the manuscript.

5. Additional clarifications

We would like to sincerely thank the reviewer and the editorial team for their insightful and constructive feedback. The comments provided have significantly contributed to the development and refinement of our manuscript. We have addressed all suggestions to the best of our ability, and we genuinely hope that the revised version meets your expectations.