Emotional Revitalization of Traditional Cultural Colors: Color Customization Based on the PAD Model and Interactive Genetic Algorithm—Taking Liao and Jin Dynasty Silk as Examples

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript introduces a compelling framework for the assessment of subjective criteria in interactive optimization algorithms. The presented results are rigorously validated, and the methodological explanations are sufficiently detailed to ensure replicability. Several minor editorial corrections are recommended, including a LaTeX error ("endenumerate") on page 11, line 361. Furthermore, the phrasing on page 13, line 414—"As shown in Fig6a, Fig6b–f depicts the initial population"—is ambiguous and would benefit from clarification. I wish the authors success with their ongoing research.

1. What is the main question addressed by the research?

The primary research question is how to effectively bridge the gap between the abstract emotional semantics of traditional cultural heritage and modern, personalized color design. The study investigates whether integrating the PAD emotional model with an Interactive Genetic Algorithm can create a system that not only responds to individual user preferences but also accurately quantifies, targets, and regenerates the specific cultural-emotional values embedded in historical artifacts.

2. Do you consider the topic original or relevant to the field? Does it address a specific gap in the field?

Yes, the topic is highly original and relevant.

While traditional genetic algorithms optimize quantifiable functions, emotions are inherently subjective and difficult to measure. The synergy between the Interactive Genetic Algorithm (IGA) and the PAD emotional model effectively bridges this gap, providing a robust methodological framework for researchers in similar contexts where quantifying subjective criteria is essential.

3. What does it add to the subject area compared with other published material?

This work adds several distinct contributions:

• A Novel "Emotional Navigator" for IGA: The core innovation is the transformation of the PAD model from a passive measurement tool into an active "emotional navigator" and fitness function within the IGA.

• Cultural-Emotional Quantification Lexicon: The development of a mapping table between Liao-Jin cultural keywords (e.g., "Bold," "Victory") and their PAD coordinates is a significant contribution.

4. What specific improvements should the authors consider regarding the methodology?

The methodology is robust, well-structured, and appropriately validated. I have no further suggestions for its improvement.

5. Are the conclusions consistent with the evidence and arguments presented?

Yes, the conclusions are strongly supported by the evidence.

6. Are the references appropriate?

Yes, the references are comprehensive and appropriate.

7. Any additional comments on the tables and figures.

Minor Typo: There is a minor formatting issue in Section 3.1 where teCheck appears to be a placeholder or error in the text describing the flowchart. The command \endenumerate is also present at the end of Section 3.3.2, which is a LaTeX error and should be removed.

Author Response

Comments 1: Several minor editorial corrections are recommended, including a LaTeX error ("endenumerate") on page 11, line 361. 

Response 1: Thank you for pointing this out. We agree with this comment. Therefore, we have corrected the LaTeX formatting error at the original line 361.

 

Comments 2: Furthermore, the phrasing on page 13, line 414—"As shown in Fig6a, Fig6b–f depicts the initial population"—is ambiguous and would benefit from clarification. 

Response 2: Thank you for this suggestion. We agree that the original wording was ambiguous. Therefore, we have rephrased the sentence for clarity. The revised text can be found at line 435 (page 14) of the revised manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents an original and methodologically sound study that integrates the PAD emotional model with an Interactive Genetic Algorithm (IGA) to achieve emotion-driven color customization for cultural heritage design. Using Liao–Jin dynasty silk as a case study, the research successfully bridges emotional computing, cultural semantics, and intelligent design — a rare and innovative approach within applied sciences and digital humanities.

Your work is conceptually strong and demonstrates clear academic value. The overall structure, clarity of objectives, and relevance of the literature are commendable. The integration of quantitative and qualitative validation (expert evaluation, fuzzy analysis, and consumer surveys) strengthens the reliability of your results. The contribution aligns well with current trends in affective computing, human–computer interaction, and digital cultural preservation.

However, before publication, a few minor revisions are recommended to enhance quality and readability:

Language and style. The English is generally good, but some expressions require polishing for fluency and precision (e.g., “the dignifieds,” “dignified dignity,” “First bullet;”).Revise sentence structure for smoother flow and consistency of verb tenses. A light professional language edit is recommended.

Methodological clarity. Include pseudo-code or a concise algorithmic flow diagram summarizing the PAD-IGA optimization process to improve reproducibility.
Provide justification for parameter settings (e.g., λ = 5, population size = 25, mutation and crossover rates). Briefly explain how statistical significance or confidence was assessed for the reported fitness improvements (32.6% and 10.76%).

Figures and tables. Ensure all figures are numbered sequentially and captions fully describe the contents (especially Figures 6–9). Improve image resolution and label axes in graphs showing fitness and convergence. Confirm tables follow MDPI formatting and include units or definitions where appropriate.

Discussion and comparison. Strengthen the comparison between the PAD-IGA model and other emotion-based design or AI-driven color restoration methods (e.g., GANs, valence–arousal models). Briefly discuss potential limitations such as cultural bias or inter-rater variability in emotional scoring.

References. Ensure all DOIs and journal names follow MDPI’s reference style. Maintain consistent formatting (italics for journal titles, commas before volumes).

Author Response

Comments 1: he English is generally good, but some expressions require polishing for fluency and precision (e.g., “the dignifieds,” “dignified dignity,” “First bullet;”).

Response 1: Thank you for this positive feedback and valuable suggestions. We agree that the language could be further polished for better fluency and precision. Therefore, we have thoroughly revised the manuscript accordingly, which includes rephrasing or removing the specific problematic expressions (e.g., "the dignifieds", "dignified dignity", "First bullet;") and refining sentence structures and verb tenses throughout the text. The changes can be found in multiple locations, including but not limited to lines 64, 148, 298, and 124, as a result of a light professional language editing process.

 

Comments 2: Include pseudo-code or a concise algorithmic flow diagram summarizing the PAD-IGA optimization process to improve reproducibility. Provide justification for parameter settings (e.g., λ = 5, population size = 25, mutation and crossover rates). Briefly explain how statistical significance or confidence was assessed for the reported fitness improvements.

Response 2: Thank you for these constructive suggestions. We agree with these comments. Therefore, we have supplemented the pseudocode and detailed justifications for parameter settings at line 367 (page 10), and added the explanation regarding the assessment of statistical significance for the reported fitness improvements at line 471 (page 16) in the revised manuscript.

 

Comments 3: Ensure all figures are numbered sequentially and captions fully describe the contents (especially Figures 6–9). Improve image resolution and label axes in graphs showing fitness and convergence. Confirm tables follow MDPI formatting and include units or definitions where appropriate.

Response 3: Thank you for these detailed suggestions. We agree with the comments. Therefore, we have carefully checked and revised all figures and tables. Specifically, we have supplemented Figure 9, added complete descriptions for Figures 6–9, improved image resolution, ensured all axes in fitness and convergence plots are properly labeled, and confirmed all tables adhere to MDPI formatting with appropriate units or definitions. These revisions can be found in the revised manuscript at line 435 (page 14) and line 461 (page 16).

 

Comments 4: Strengthen the comparison between the PAD-IGA model and other emotion-based design or AI-driven color restoration methods (e.g., GANs, valence–arousal models). Briefly discuss potential limitations such as cultural bias or inter-rater variability in emotional scoring.

Response 4: Thank you for these insightful comments. We agree with the suggestions. Therefore, we have strengthened the comparative discussion between the PAD-IGA model and other emotion-based or AI-driven color restoration methods (such as GANs and valence-arousal models). Additionally, we have included a brief discussion on potential limitations. These supplements can be found at line 572 (page 18) and line 579 (page 19) in the revised manuscript.

 

Comments 5: References. Ensure all DOIs and journal names follow MDPI’s reference style. Maintain consistent formatting (italics for journal titles, commas before volumes).

Response 5: Thank you for pointing this out. We agree with the comment. Therefore, we have carefully checked and unified the reference format throughout the manuscript to ensure full compliance with MDPI's style guide. 

Reviewer 3 Report

Comments and Suggestions for Authors

Title:

Emotional Revitalization of Traditional Cultural Colors: Color Customization Based on the PAD Model and Interactive Genetic Algorithm - Taking Liao and Jin Dynasty Silk as Examples

Understanding the topic:

This manuscript elaborates a digital regeneration pathway for traditional cultural colors,
evolving from “form-color restoration” to “emotional awakening.” To achieve progress the authors constructed an emotional color customization system through the integration of Interactive Genetic Algorithms (IGA) with the PAD emotional model. 

Actual experiment was conducted with 48 volunteer evaluators using Liao-Jin silk templates. According to authors, experimented results demonstrated that in compared to others methods like traditional IGA, this method achieved significant improvements in emotional matching accuracy. More precisely, average fitness is increased by 32.60%, maximum fitness rose by 10.76%, and consequently the spiritual essence of Liao-Jin culture was more effectively translated into color schemes that evoke positive user emotions.

Field: Computer vision, recognition patterns and algorithms

Under the field - specialty: “form-color restoration” and “emotional and spiritual regeneration” 

Clarification:

This research gives an innovative solution for cultural heritage digitization, advancing from “form-color restoration” to “emotional and spiritual regeneration.” It also provides a viable approach for intelligent emotional design in fields such as apparel design, cultural creativity, and digital cultural heritage preservation.

Review:

The topic is well described through emotional resonance with consumers and evoking aesthetic experience throw consequently influence purchasing decisions and brand loyalty. The relationship is correctly described: collaborative decision-making between man and machine and the architecture of interactive genetic algorithms (IGA). The historical sequence and development of PAD model methods is also elaborated. The authors clearly indicated three main contributions of this study.

The Interactive Genetic Algorithm (IGA) method is correctly described and equipped.

Experiment results:

The PAD values, SAM scale, and color optimization process are shown in the table and diagram. Validation of the proposed PAD-IGA color matching method was achieved thanks to a set of 48 volunteers. Designed color scheme and satisfactory color scheme are presented in Figures 6 and 7, also convergence trend (Fitness Change Trend) in Figure 8. 

Results:

According to the color scheme evaluation scoring shown in Table 2 , the authors demonstrated that the PAD-IGA method not only achieves emotion-driven optimization at the technical level, but also shows strong applicability in cultural semantic representation and market acceptance.

Conclusion:

According to the presented data and the proposed scientific experimental method, a new computational framework for emotional color design is revealed and methodological support is offered for the contemporary interpretation of traditional cultural elements.

A list of references is sufficient.

 

Author Response

Comments : Are the conclusions supported by the results? Must be improved.

Response : Thank you for this critical comment. We agree that the conclusions must be firmly grounded in the experimental results. Therefore, we have thoroughly revised the conclusion section to strengthen the logical connection between the findings and our final arguments. The key revisions include: 1) explicitly linking the fitness improvements to the method's effectiveness; 2) incorporating the statistical assessment of these results; 3) directly connecting the evaluation scores from Table 2 to our claims about cultural applicability. These revisions can be found at line 367 (page 10), line 461 (page 16), and line 572 (page 18) in the revised manuscript.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors present an algorithm based on genetic programming to map emotions to colors. I found this investigation interesting, although I am skeptical about the idea that all emotions can be quantified or computed. Here are my comments:

1. How many generations did the algorithm require to achieve desirable results?
2. How do the authors control the number of mutations for the different functions involved?
3. I didn’t understand what regularization the authors implemented to prevent overfitting.
4. How challenging is it to map emotions to colors for the entire population of China? Would this challenge change if we considered different historical time frames?

Author Response

Comments 1:  How many generations did the algorithm require to achieve desirable results?

Response 1: Thank you for this question regarding the convergence efficiency of our algorithm. We have added clarifications in the manuscript. As specified in the Methods section (line 338 page 10), an objective convergence criterion was defined, and a maximum generation limit of G_max = 15 was set. Figure 8 (line 447 page 15) and Figure 9 (line 461 page 16) show that both the population distribution and fitness values stabilized around the 13th generation, indicating that the algorithm typically converges within 13–15 generations. 

 

Comments 2: How do the authors control the number of mutations for the different functions involved?

Response 2: Thank you for raising this question about the algorithm's implementation details. In our PAD-IGA system, control over the “number of mutations” primarily manifests at two levels: the probability of mutation occurrence and the magnitude of mutation operations. We employ a precise mechanism to finely regulate this process, ensuring a balance between maintaining population diversity and preserving desirable traits.
The specific control mechanisms are as follows: (1) Fixed mutation probability. We set a global mutation probability (p_m = 0.3). When generating a new generation population, for each color scheme individual in the offspring population, the system independently decides with a 30% probability whether to apply a mutation operation to it. (2) Controlling mutation intensity. When an individual's color scheme is selected for mutation, the algorithm does not randomly generate entirely new colors. Instead, it applies small, bounded perturbations to each existing color. Specifically, for each color gene (i.e., an RGB vector) in the scheme, we perform the following operation on its R, G, and B channels: generate a random change Δ uniformly distributed within the range [-MAX_MUTATION_CHANGE, MAX_MUTATION_CHANGE]. In our experiments, MAX_MUTATION_CHANGE is set to 0.3. The new channel value is calculated as: new_value = old_value * (1 + Δ). Finally, the new channel value is clipped to the valid range [0, 255].

In summary, this mechanism ensures controllable mutation intensity and uniformity in functional processing. The parameter MAX_MUTATION_CHANGE directly controls the maximum relative amplitude of color changes, enabling mutations to generate new explorations without completely disrupting the existing harmonious color structure of the parent generation. Currently, we apply a uniform mutation probability and amplitude to all color genes. This is because our goal is the overall emotional expression of the color scheme, rather than emphasizing the functionality of any specific color. This uniform strategy effectively drives evolution while maintaining simplicity and avoiding over-parameterization.

 

Comments 3: I didn’t understand what regularization the authors implemented to prevent overfitting.

Response 3: We sincerely thank the reviewer for raising this crucial question. In the context of Genetic Algorithms and Interactive Genetic Algorithms, the issue referred to as "overfitting" is more accurately termed "premature convergence" – where the algorithm converges too early to a local optimum and fails to explore potentially better global solutions. We fully agree on the importance of preventing this issue.
Although the term "regularization" is not directly used in evolutionary computation, our PAD-IGA system incorporates multiple mechanisms that achieve equivalent or even stronger functions to maintain population diversity and exploration capability. Together, these mechanisms effectively prevent premature convergence: 1) Elitist Strategy: In each generation, we directly preserve a small number of elite individuals with the highest fitness into the next generation. This prevents the loss of high-quality solutions already discovered due to random operations. It stabilizes the evolutionary process by ensuring that progress is not derailed by random fluctuations in a single generation, analogous to periodically saving the best-performing model on a validation set during training. 2) Roulette Wheel Selection: We introduce stochasticity to smooth the optimization path. This injected randomness acts as a form of noise, preventing the algorithm from rigidly following the currently perceived optimal path. This mechanism is somewhat similar to adding random perturbations to the optimization objective. 3) Mutation Operation: As detailed in our response to the question about controlling the number of mutations, we apply bounded, small-magnitude color perturbations to offspring individuals with a probability of p_m=0.3. Mutation continuously injects new genetic material into the population, persistently exploring new regions of the color space and thereby breaking the algorithm's fixed mindset or entrenchment in suboptimal solutions. 4) The User as a Dynamic and Noisy Fitness Function: In traditional machine learning, overfitting occurs when a model over-adapts to noise in a static training dataset. In our IGA, the "training data" consists of real-time, subjective emotional feedback from the user. Factors such as user fatigue, mood fluctuations, and subtle inconsistencies in evaluation criteria inherently introduce noise into the fitness function. This noise makes it difficult for the algorithm to over-optimize toward any single, absolute, and static target point. It compels the algorithm to find a relatively robust solution—one that maintains high fitness even when user ratings exhibit minor variations. This represents a unique and inherent anti-overfitting characteristic of the IGA framework.
In summary, we have not overlooked the issue of overfitting. Instead, we address it through a comprehensive set of systematic strategies specifically designed for interactive evolution. These measures ensure that our PAD-IGA can robustly explore the emotional color space and generate culturally and emotionally valuable color schemes that are not overspecialized.

 

Comments 4: How challenging is it to map emotions to colors for the entire population of China? Would this challenge change if we considered different historical time frames?

Response 4: Thank you for raising this crucial question. The reviewers correctly point out that establishing a universal “emotion-color” mapping model applicable to the entire Chinese population across different historical periods is extremely challenging. This difficulty stems from China's vast geographical, ethnic, and cultural diversity, as well as the fundamental shifts in aesthetic paradigms and color symbolism throughout different eras. For instance, the opulent splendor of the Tang Dynasty and the understated elegance of the Song Dynasty expressed “nobility” in starkly different ways. The core value of this study lies in proposing a transferable computational framework. The method's high context-specificity—relying on an emotion PAD dictionary constructed within the Liao-Jin cultural context—is precisely what enables its precise “emotion restoration.” When applying this framework to new historical periods or specific cultural groups, the core PAD-IGA algorithm remains unchanged. Instead, it is achieved by replacing the existing emotional dictionary with a new one based on domain-expert knowledge and historical texts. Regarding the challenges and future directions, we have supplemented the content as discussed. It can now be found at line 592 (page 19).