Cosmetic quality grading method based on dynamic weight optimization

Abstract

The application discloses a cosmetic quality grading method based on dynamic weight optimization, comprising the following steps: step one, detection and result scoring; step two, weight matrix construction; step three, comprehensive score calculation; step four, synergistic gain correction; and step five, quality grade division. The application proposes a hybrid grading model of "basic threshold + dynamic weight", and verifies the effectiveness and operability of the grading model through application examples of representative categories. The application aims to provide technical support for high-quality development of the industry and help consumers make more intelligent purchase decisions.

Description

Technical Field

[0001] This invention relates to a cosmetic quality grading method, and more particularly to a cosmetic quality grading method based on dynamic weight optimization. Background Technology

[0002] Consumer product quality grading is a technical means of classifying product quality into different levels through standardized evaluation. Its core value lies in overcoming the dilemma of "information asymmetry"—providing consumers with comparable quality signals and clarifying the direction for quality improvement for enterprises. Consumer product quality grading has entered a stage of systematic development, but the implementation effects vary significantly across different sectors, and personalized consumer products such as cosmetics still lack unified grading standards.

[0003] As a special category of products that come into direct contact with the human body, cosmetics face three major challenges in quality evaluation: first, the discrepancy between efficacy claims and actual effects (e.g., the tyrosinase inhibition rate of whitening products is inconsistent with clinical results, making it difficult for consumers to discern the true efficacy); second, the dynamic changes in safety (e.g., excessive microorganisms caused by preservative degradation may pose safety hazards during use); and third, individual differences in user experience (e.g., the different absorption efficiencies of moisturizers for different skin types, making it difficult for a single evaluation standard to cover diverse needs). Existing cosmetic quality evaluation research mostly focuses on measuring safety risks, lacking a systematic framework and grading model. Summary of the Invention

[0004] Current regulations primarily focus on basic safety requirements, making it difficult to guide enterprises towards higher quality standards. A tiered management system can help achieve precise policy implementation. Faced with a dazzling array of cosmetics, ordinary consumers struggle to accurately judge the actual quality of products. This invention, based on a dynamic weight optimization-based cosmetic quality grading method, can provide intuitive and reliable quality level information to guide consumers' purchasing decisions.

[0005] A cosmetic quality grading method based on dynamic weight optimization includes the following steps: Step 1: Testing and Result Scoring: Indicators for four dimensions of cosmetics—safety, efficacy, consistency, and user experience—are selected and tested to obtain measured values. Scores are assigned based on the test results, and the scores ψ for each of the four dimensions are calculated. S ψ E ψ C and ψ U ; Step 2, Weight Matrix Construction: By combining basic weight settings with pattern-based adjustments, the weight ratio of each dimension is allocated differently for different application scenarios such as supervision, testing, and R&D. At the same time, pattern enhancement coefficients are used to strengthen the core objective orientation and dynamically adjust the weights to form a weight method that takes into account both universality and specificity. Step 3: Calculate the overall score: Based on the scores of the four dimensions and the dynamically adjusted weights, calculate the overall score; Step 4: Cooperative gain correction: When both efficacy and user experience reach high standards, that is, ψ E ≥ 0.80 and ψ U When the score is ≥ 0.85, the product's final score is... λ is the gain score, calculated as follows:

[0006] Step 5: Quality Grade Classification A comprehensive score greater than 95 points is AAAAA Excellent level; a comprehensive score of 85-95 points is AAAA Good level; a comprehensive score of 75-85 points is AAA Preferred level; a comprehensive score of 65-75 points is AA Standard level; a comprehensive score of 60-64 points is A Basic Pass level; a comprehensive score less than 60 points or in step one ψ S When the value is less than 0.6, it is considered as not meeting the standard.

[0007] The cosmetic quality grading method based on dynamic weight optimization described in this invention includes step two as follows: The basic weights for safety, efficacy, consistency, and user experience are 0.30, 0.25, 0.25, and 0.20, respectively. The basic weights are dynamically adjusted in different application scenarios: Based on the core objectives of different application scenarios, the weights of key dimensions are increased accordingly. A scenario-specific enhancement coefficient β is introduced to amplify the weights of the core dimensions, strengthening their actual impact. The final applied weight A for the core dimensions is... x The calculation formula is as follows:

[0008] in, β： A unique enhancement factor, ranging from 1.0 to 1.5; At the same time, the weights of non-core dimensions are proportionally reduced to ensure that the sum of the weights of all dimensions remains 1 after the adjustment. The specific formula is as follows:

[0009] Among them: A 0 The original weights of the core dimensions; A: The original weights of non-core dimensions; A X Application weights of core dimensions; A X ′: Application weight of non-core dimensions.

[0010] The cosmetic quality grading method based on dynamic weight optimization described in this invention, wherein: Under the regulatory model, the focus is on risk prevention and control. β The score is 1.5, with the core dimensions being safety and consistency. After calculation, the basic weights for safety, efficacy, consistency, and user experience are 0.45, 0.097, 0.375, and 0.078, respectively. In the testing mode, the focus is on objective data verification, with efficacy and consistency as the core dimensions. β The base weights for safety, efficacy, consistency, and user experience are 1.2, 0.24, 0.30, 0.30, and 0.16, respectively. Under the R&D model, innovation and market value are the guiding principles, with efficacy and user experience as the core dimensions. β The base weights for safety, efficacy, consistency, and user experience are 1.3, and after calculation, they are 0.226, 0.325, 0.189, and 0.26, respectively.

[0011] The cosmetic quality grading method based on dynamic weight optimization described in this invention, wherein: The calculation method for the overall score in step three is as follows:

[0012] Among them: A S A E A C A U The application weights are dynamically adjusted; ψ S , ψ E , ψ C , ψ U Scores are given for each indicator.

[0013] The cosmetic quality grading method based on dynamic weight optimization described in this invention, wherein: The safety indicators and their scores in Step One are as follows:

[0014] Security score ψ S The calculation method is as follows: Step 1: Check if there are any "veto" items; If any one of the following tests—prohibited substance detection, restricted substance control, total bacterial count, mold and yeast detection, pathogenic bacteria detection, preservative efficacy evaluation, skin patch test, and eye irritation test—scores 0, then ψ S = 0, which means "one vote veto", and no further calculation is required; Step 2: If there is no "veto" item, then take a weighted average of the scores of all secondary indicators to obtain ψ.S Score; If ψ after calculation S If the score is less than 0.6, it will be directly rated as substandard and no further evaluation will be conducted.

[0015] The cosmetic quality grading method based on dynamic weight optimization described in this invention, wherein: The efficacy indicators and scoring rules are as follows:

[0016] Efficacy score ψ E Calculation method: Based on the product's claimed efficacy, provide at least one type of valid evidence, either in humans or in vitro, and select 1-2 core indicators for scoring. If multiple types of evidence are provided, the highest score is taken as the single efficacy score. If the product claims multiple efficacy, the corresponding total score is calculated. This is the arithmetic mean of the scores of all claimed power components.

[0017] The cosmetic quality grading method based on dynamic weight optimization described in this invention, wherein: The consistency indicators and scoring rules are as follows:

[0018] Consistency score ψ C The calculation method is the arithmetic mean of each indicator with equal weight.

[0019] The cosmetic quality grading method based on dynamic weight optimization described in this invention, wherein: The experiential metrics and scoring rules are as follows:

[0020] User experience score ψ U The calculation method is: the arithmetic mean of each indicator with equal weight.

[0021] This invention, based on dynamic weight optimization, differs from existing technologies in that it addresses the current cosmetics market's problems of inconsistent quality, exaggerated efficacy claims, and the urgent need for a scientific grading system. By combining the quality characteristics of cosmetics, this invention designs and constructs a multi-level indicator system based on the four dimensions of "safety, efficacy, consistency, and user experience." It proposes a hybrid grading model of "basic threshold + dynamic weights," and verifies the effectiveness and operability of the grading model through application examples in representative product categories. The aim is to provide technical support for the high-quality development of the industry and help consumers make more informed purchasing decisions.

[0022] The following detailed description of the cosmetic quality grading method based on dynamic weight optimization of the present invention will be further illustrated with specific embodiments. Detailed Implementation

[0023] Construction of a cosmetic quality grading index system: 1. Design of a hierarchical indicator system Combining the characteristics of cosmetics in terms of "safety, efficacy, consistency, and experience," and based on the principles of "safety first, efficacy as the core, consistency as the guarantee, and experience as the supplement," a four-dimensional indicator system for grading the quality of cosmetics is constructed.

[0024] 1.1 Safety Indicators and Scoring Rules

[0025] Safety is a mandatory requirement for cosmetics, and the selection of indicators follows the principle of "comprehensive risk control," covering three levels: ingredient safety, microbial control, and irritation evaluation. Specifically, the detection of prohibited substances follows the requirements of the "Cosmetic Safety Technical Specifications" (2015 edition), adopting a "zero tolerance" standard (i.e., no detection allowed); the control of restricted substances is differentiated according to "residue / rinse-off" categories, adhering to national standards; and the toxicological evaluation covers safety test results such as skin irritation and eye irritation, ensuring the objectivity of the evaluation.

[0026] Table 1 Safety Index Parameter Table

[0027] Security score ψ S The calculation rules are as follows: Step 1: Check if there is a "veto" item. If any secondary indicator score is 0 (e.g., detection of prohibited substances, excessive levels of restricted substances, detection of pathogenic bacteria, serious exceedance of microbial limits, etc.), then ψ S =0, meaning "one vote veto", no further calculation is needed. Step 2: If there is no veto, then calculate a weighted average of the scores for all secondary indicators.

[0028] Among them, the methods in the "Cosmetic Safety Technical Specifications" (2015 edition) were used to test for prohibited substances, restricted substances, total bacterial count, molds and yeasts, pathogenic bacteria, and preservative efficacy. Skin patch test: Methodology: The test was conducted using the "Human Skin Patch Test" method as per the "Cosmetic Safety Technical Specifications" (2015 edition). Key points of the method: Applicable to cosmetic products; observe and record results according to the "Skin Reaction Grading Standard for Closed / Open Patch Tests". In human closed patch tests, if more than 5 out of 30 subjects experience Grade 1 adverse skin reaction (excluding 5), or more than 2 out of 30 subjects experience Grade 2 adverse skin reaction (for deodorant patch tests, more than 10 out of 30 subjects experience Grade 1 adverse skin reaction, or more than 5 out of 30 subjects experience Grade 2 adverse skin reaction), or if any one subject experiences Grade 3 or higher adverse skin reaction, the test substance is considered to have an adverse skin reaction in humans. (Refer to "Cosmetic Hygiene Standards (2007 Edition)") Eye stimulation test: Method based on: SN / T 2329-2009 "Cosmetic Eye Irritation / Corrosivity Test with Chicken Embryo Villi Allantoic Membrane" Key points of the method: Applicable to cosmetics and cosmetic raw materials; results can be evaluated based on IS value (stimulus score) or ES (endpoint score).

[0029] 1.2 Efficacy Indicators and Scoring Rules

[0030] Efficacy is the core dimension of cosmetic quality grading, and its indicator settings need to consider multiple characteristics: it must cover the time difference between immediate and long-lasting effects, the evidence levels for apparent improvement and mechanism verification, and adapt to individual differences in different skin types. This study constructs a three-tiered system of "efficacy claim—evidence type—quantitative indicators" around 16 types of efficacy claims in cosmetics, including hair loss prevention, whitening and spot removal, and sun protection. The indicator design follows the "dual-track verification" principle: human trials focus on the actual efficacy presentation (such as intuitive parameters like transepidermal water loss and hair density), while in vitro experiments emphasize the analysis of biological mechanisms (such as molecular-level indicators like collagen gene expression levels). All efficacy indicators must simultaneously meet statistical significance (P<0.05, α=0.05), providing an operational basis for quality grading of cosmetics with various efficacy claims.

[0031] Table 2 Efficacy Indicator Parameter Table

[0032] Efficacy score ψ E The calculation rules are as follows: Based on the product's claimed efficacy, at least one type of valid evidence (human / in vitro) must be provided, and 1-2 core indicators must be selected for scoring. If multiple types of evidence are provided, the highest score will be used as the score for a single efficacy claim. If the product claims multiple efficacy claims, the corresponding total score will be the arithmetic mean of the scores for all claimed efficacy sub-claims.

[0033] In this invention, the methods used to obtain the results of primary endpoint indicators, secondary endpoint indicators, subjective evaluation indicators, mechanistic indicators, and physicochemical indicators are as follows: Taking repair effects as an example: Human body legal basis: T / CAFFCI 66-2023 Test Methods for Repair Efficacy of Cosmetics (First Law Human Efficacy Evaluation Test Method); Key indicator: Transdermal water loss rate; Secondary indicators: stratum corneum moisture content, skin redness; Subjective assessment: Visual assessment 1 (dryness, scaling), Visual assessment 2 (redness), Visual assessment 3 (smoothness); In vitro testing was conducted according to: T / ZHCA 020-2022 Method for Testing the Migration Ability of Human Fibroblasts in In Vitro for Testing the Repairing Efficacy of Cosmetics; Mechanistic indicators: Using image analysis software (such as ImageJ) to measure scratch width or area, and calculate migration rate or healing rate. Rate, to assess cell migration ability; In vitro testing was conducted according to: T / CAFFCI 66-2023 Test Method for Repairing Efficacy of Cosmetics (Second Method for Evaluating the Efficacy of In Vitro Hair). Physicochemical properties: tensile strength, combing work, friction work.

[0034] Taking anti-wrinkle effects as an example: Human body method basis: Test method for anti-wrinkle efficacy of cosmetics (Group standard of the Fragrance and Cosmetics Association, to be released soon) T / ZHCA006-2019 Test method for anti-wrinkle efficacy of cosmetics; Key indicators: Ra and Rz obtained from skin 3D optical imaging test analysis, wrinkle area, wrinkle area percentage, wrinkle volume, and average wrinkle depth; Secondary indicators: R3 and R5 obtained from skin 2D texture imaging test, wrinkle area and wrinkle area percentage obtained from standard image shooting test analysis; Subjective indicators: Wrinkle levels are obtained through visual assessment based on the wrinkle level assessment scale and wrinkle level standard atlas. In vitro method basis:

[0035] Mechanistic indicators: zebrafish juvenile caudal fin wrinkling inhibition rate, in vitro fibroblast type I collagen content, in vitro keratinocyte reactive oxygen species (ROS) inhibition rate, fibroblast advanced glycation end product carboxymethyl lysine content, elastase inhibition, anti-non-enzymatic glycation ability, free radical scavenging rate, etc.

[0036] Taking acne-removing effects as an example

[0037] Human testing standards: T / CAFFCI 67-2023 Human testing methods for the acne-removing efficacy of cosmetics; T / CNMIA 0012-2020 Clinical evaluation standards for acne-removing functional skin care products. Key indicator: Acne papule count; Secondary indicators: image analysis parameters, sebum, and stratum corneum moisture content; Subjective indicators: overall performance score, subject self-assessment; In vitro methods are based on: T / CNMIA 0010-2020 Safety / Efficacy Evaluation Standard for Acne-Removing Functional Skin Care Products; T / LNPC 018-2024 Test Method for Evaluating the Acne-Removing Efficacy of Cosmetic Raw Materials; T / QLMZ 8-2023 Evaluation Method for the Acne-Removing Efficacy of Cosmetics Based on Zebrafish Model; Mechanistic indicators: inhibition of Propionibacterium acnes biofilm activity, effect of IL-8 expression, inhibition of Propionibacterium acnes, rat auricular acne model.

[0038] 1.3 Quality Consistency Indicators and Scoring Rules

[0039] Consistency in quality is a key dimension in cosmetic production and quality control, concerning the stable presentation of core qualities such as safety, efficacy, and user experience across different batches, production cycles, and even environments. Poor consistency in quality not only weakens the consumer experience but may also lead to safety risks. This study focuses on key nodes throughout the entire cosmetic production process, selecting core indicators that reflect the stability of product quality. Ingredient consistency reflects batch-to-batch variations in active ingredient content, ensuring consistent product quality. Sensory stability includes the stability of sensory characteristics such as color, odor, and texture, impacting the consumer experience. Packaging integrity encompasses packaging sealing, material safety, and labeling compliance, ensuring product quality during storage and transportation.

[0040] Table 3 Quality Consistency Index Parameter Table

[0041] The calculation rules for the quality consistency score ψC are as follows: Unless otherwise specified, all secondary quality indicators are considered to be of equal importance, and the total score is the arithmetic mean of each indicator with equal weight.

[0042] in: Content of key ingredients: based on the testing methods specified in the national or industry standards corresponding to the product, or based on the methods in the "Cosmetic Safety Technical Specifications" (2015 edition).

[0043] For example, the method for using sunscreen agents is based on: the "Cosmetic Safety Technical Specifications".

[0044] GB / T 30933-2014 Determination of diethylaminohydroxybenzoyl benzoate, a sunscreen agent, in cosmetics - High performance liquid chromatography; GB / T 35916-2018 Determination of 16 permitted sunscreens and 8 other ultraviolet-absorbing substances in cosmetics by high performance liquid chromatography; GB / T 30933-2014 Determination of diethylaminohydroxybenzoyl benzoate, a sunscreen agent, in cosmetics - High performance liquid chromatography; pH value: 《Cosmetic Safety Technical Specifications》, GB / T 13531.1-2008 General Test Methods for Cosmetics - Determination of pH value; Color change, odor stability, and sealing performance: T / SHFCA 002-2021 Guidelines for Cosmetic Stability Testing; Migration of hazardous substances: GB 31604.1-2023 National Food Safety Standard General Rules for Migration Test of Food Contact Materials and Articles.

[0045] 1.4 User Experience Metrics and Scoring Rules

[0046] User experience is also a key dimension for measuring market acceptance and user loyalty in the cosmetics industry, focusing on subjective feelings and overall satisfaction during actual product use. This indicator aims to quantify consumers' intuitive perceptions and preferences regarding product appearance, texture, skin feel, and long-term effects, effectively compensating for individual differences and emotional feedback that are difficult to capture with traditional objective indicators. This study constructs a user experience evaluation system from three levels: preference, usability, and long-term satisfaction. Through a combination of standardized questionnaires and structured interviews, subjective feelings are transformed into quantifiable data, providing important consumer-side evidence to support product quality grading.

[0047] Table 4 User Experience Metrics Parameter Table

[0048] User experience score ψ U Calculation rules: Unless otherwise specified, all secondary user experience indicators are considered equally important, and the total score is the arithmetic average of all indicators with equal weight.

[0049] Emotional preferences and aesthetic preferences are usually assessed using subjective evaluation methods. The main approach involves developing scales to collect raw data and then using descriptive statistics or other testing methods for data analysis.

[0050] Methodological basis: T / GDCA 003-2020 General Rules for Sensory Evaluation of Cosmetics; GB / T 19547-2025 Sensory Analysis Methodology: Measurement Estimation Method; GB / T 43396-2023 "Guidelines for Sensory Claim Verification of Sensory Products"; Parameter values: In sensory evaluation of cosmetics, following the principles of scale design, rating levels should be set based on actual needs. The actual level reflects the degree of acceptance or recognition on the scale, not an absolute value. Taking the Likert scale as an example, respondents are typically asked to indicate their level of agreement with the statement on the item, or any form of subjective or objective evaluation. Five response levels are usually used, but in practice, options of 7, 9, 10, and 15 levels can all be applied, mainly based on the needs of the item; the more levels, the more detailed the information obtained.

[0051] Questionnaire: Questionnaires can be designed based on project needs or selected from publicly available resources such as literature. This study uses a questionnaire, which was developed specifically for the project. Basic principles of the questionnaire: 1. Minimum 2 weeks; 2. Can be conducted online or offline; 3. Minimum 30 participants.

[0052] 2. Selection of Cosmetic Quality Grading Methods

[0053] This invention proposes a hybrid hierarchical path of "basic threshold + dynamic weight".

[0054] 2.1 Four-dimensional indicator system

[0055] Based on the aforementioned indicator system and scoring rules, values ​​were assigned by referring to Tables 1 to 4 using measured data. Combining the veto mechanism and the equal-weighted averaging method, the safety ψ was calculated respectively. S Efficacy E Consistency of quality C and user experience ψ U The scores across the four dimensions provide a quantitative basis for subsequent comprehensive quality evaluation.

[0056] 2.2 Construction of Weight Matrix

[0057] By combining basic weight settings with pattern-based adjustments, the weight ratio of each dimension is allocated differently for different application scenarios such as supervision, testing, and R&D. At the same time, a pattern enhancement coefficient is used to strengthen the core objective orientation, forming a weight system that takes into account both universality and specificity.

[0058] (1) Basic weights

[0059] The Delphi method is used to determine the benchmark proportion of each dimension in a general evaluation scenario, reflecting the system setting principle of "safety first, efficacy as the core, consistency as the guarantee, and experience as a supplement".

[0060] Safety (S): Weight 0.30. As the baseline for cosmetic quality, it covers direct risk indicators such as prohibited substances and microbial control, and is the basic guarantee for all scenarios, hence it is set with the highest basic weight.

[0061] Efficacy (E): Weight 0.25. Reflects the core value of the product (such as claimed effects like moisturizing and whitening), and is a key basis for consumer decision-making. Together with experiential factors, it constitutes the product's competitiveness.

[0062] Consistency (C): Weight 0.25. Ensuring consistent product quality throughout the entire lifecycle of production, storage, and use (such as batch-to-batch differences in active ingredients and pH stability) is a crucial support for long-term quality.

[0063] User Experience (U): Weight 0.20. This factor encompasses user comfort (ease of use, stickiness) and long-term satisfaction (repurchase intention, recommendation rate), directly impacting user acceptance and market feedback.

[0064] (2) Dynamic adjustment of weights

[0065] Based on the core objectives of different application scenarios, the weights of key dimensions are increased accordingly. A scenario-specific enhancement coefficient β is introduced to amplify the weights of core dimensions and strengthen their actual impact. The final applied weight A for each core dimension is... x The calculation formula is as follows:

[0066] in, β： Exclusive enhancement factor (value range 1.0-1.5).

[0067] At the same time, the weights of non-core dimensions are proportionally reduced (ensuring that the sum of the weights of all dimensions remains 1 after the adjustment), as shown in the following formula:

[0068] Where A 0 The original weights of the core dimensions; A: The original weights of non-core dimensions; A X Application weights of core dimensions; A X ′: Application weight of non-core dimensions.

[0069] (3) Example of adjusting logic

[0070] Regulatory model: Focuses on risk prevention and control, with safety as the core dimension (A) S ) and consistency (A CThe original core total = 0.30 + 0.25 = 0.55, and the new core total = 0.45 + 0.375 = 0.825; non-core (A) E A U After compression according to the formula, the proportion is 0.175, which matches the goal of "prioritizing the bottom line of compliance".

[0071] Testing mode: Focusing on objective data verification, with efficacy as the core dimension (A) E ) and consistency (A C The original core total = 0.25 + 0.25 = 0.50, and the new core total = 0.30 + 0.30 = 0.60; non-core (A) S A U After compression, the proportion is 0.40, which strengthens the need for "experimental data support".

[0072] Research and development model: innovation- and market-value-oriented, with efficacy as the core dimension (A) E ) and experiential (A U The original core total = 0.25 + 0.20 = 0.45, and the new core total = 0.325 + 0.26 = 0.585; non-core (A) S A C After compression, the proportion is 0.415, which aligns with the goal of "enhancing product competitiveness".

[0073] The dynamic weighting indicators are allocated as shown in the table below.

[0074] Table 5 Dynamic Weight Allocation Table

[0075] 2.3 Calculation of Overall Score

[0076] Basic Comprehensive Score

[0077] Among them: A S A E A C A U The application weights are dynamically adjusted. ψ S , ψ E , ψ C , ψ U The scores are the standardized indicators.

[0078] 2.4 Synergistic Gain Correction

[0079] Synergistic gain correction follows the principle of "dual incentive of efficacy and experience." For cosmetics that simultaneously achieve high standards in both efficacy (E) and user experience (U), a quantitative formula is used to positively correct the base score. This principle emphasizes the synergistic value of "high efficacy" and "high experience" (triggered only when E ≥ 0.80 and U ≥ 0.85), while using a non-linear function to avoid excessive influence from a single dimension, ensuring that the gain is tilted towards products with "solid efficacy and excellent experience," ultimately achieving precise incentives for high-quality products. The specific formula is as follows: When both efficacy (E) and user experience (U) reach a high standard (i.e., trigger condition: ψ) E ≥ 0.80 and ψ U ≥ 0.85); The final score of the product

[0080] Where λ is the gain score, determined by the joint level of E and U. As shown in the following equation,

[0081] For example, when ψ E =0.90、ψ U When λ = 0.90: λ = 2 × 0.9 × 0.9 × 1.075 ≈ 1.742.

[0082] 2.5 Quality Grade Classification

[0083] Quality grading is the ultimate manifestation of the cosmetic quality grading system, aiming to intuitively and clearly define the quality level of products through a comprehensive score. Based on the comprehensive score (including synergistic gain correction) calculated using the hybrid grading model of "basic threshold + dynamic weight" proposed in this paper, product quality can be divided into multiple levels from "excellent" to "substandard". This grading system strictly adheres to safety standards (i.e., a safety score below 60 is directly considered unqualified) while fully reflecting the differentiated characteristics of different quality levels, providing consumers with clear and operational grading criteria for purchasing, enterprises for quality improvement, and regulatory authorities for precise policy implementation.

[0084] Table 6. Classification Table of Hybrid Classification Model

[0085] 3. Application and Verification of Cosmetic Quality Grading

[0086] To verify the effectiveness of the hybrid grading model, five representative cosmetic categories were selected for grading tests (Table 7). The grading process strictly followed a standardized procedure: first, the indicators were preprocessed using the Delphi method (screening and standardizing core parameters); then, dynamic weight adjustments were made based on regulatory, testing, and R&D scenarios (Table 5); next, the basic comprehensive score S0 was calculated using a formula; then, synergistic gain correction λ was applied to products that met the condition of "E≥0.8 and U≥0.85"; finally, the quality level was determined based on the comprehensive score (Table 6), fully covering all aspects of the technical process.

[0087] Table 7 Comparison of Examples of Cosmetic Quality Grading

[0088] 3.1 Example Analysis

[0089] (1) Core Formula (A is the dynamic weight, and ψ is the standardized index value) Synergistic gain correction (trigger condition: ψ) E ≥ 0.80 and ψ U ≥ 0.85): S F = S0 + λ (2) Product A (Regulatory Model) Weight: A S = 0.45, A E = 0.097, A C = 0.375, A U = 0.078 S0= (0.45×0.95 + 0.097×0.9 + 0.375×0.90 + 0.078×0.83)×100=91.70 Synergistic gain: ψ U = 0.83 < 0.85, not triggered, overall score S F = 91.70 (3) Product B (Basic Model) Weight: A S = 0.30, A E = 0.25, A C = 0.25, A U = 0.20 S0= (0.30×0.83 + 0.25×0.80 + 0.25×0.70 + 0.20×0.71)×100=76.60 Synergistic gain: ψ U= 0.71 < 0.85, not triggered, overall score S F = 76.60 (4) Product C (R&D Model) Weight: A S = 0.226, A E = 0.325, A C = 0.189, A U = 0.26 S0= (0.226×0.95 + 0.325×1.0 + 0.189×0.97 + 0.26×0.94)×100=96.74 Synergistic gain: ψ E =1.0 ≥ 0.8 and ψ U = 0.94 ≥ 0.85, λ ≈ 2.15, Overall score S F =96.74 + 2.15 = 98.89 (5) Product D (Detection Mode) Weight: A S = 0.24, A E = 0.30, A C = 0.30, A U = 0.16 S0= (0.24×0.78 + 0.30×0.77 + 0.30×0.70 + 0.16×0.83)×100=76.10 Synergistic gain: ψ U = 0.83 < 0.85, not triggered, overall score S F = 76.10 3.2 Adaptability Validation of Cosmetic Quality Grading Model The system was validated from two aspects: product type coverage and scenario weight effectiveness, to ensure the scientific validity and reliability of the quality grading model.

[0090] 3.2.1 Product type coverage verification

[0091] Twelve typical cosmetic categories (covering sub-sectors such as skincare, makeup, and personal care) were selected, and five commercially available brands were randomly chosen from each category (including brands with different price points and efficacy claims), resulting in a total of 60 samples for tiered testing. No samples lacked core parameters or could not be quantified due to the specificity of dosage form or efficacy claims. The core indicators of different product categories showed significant differentiation, demonstrating that the model has broad product coverage and good category adaptability.

[0092] 3.2.2 Verification of the effectiveness of scene weights

[0093] (1) Validation of the regulatory model

[0094] Ten defective products that had previously been reported by regulatory authorities were selected as verification samples, and the regulatory model (A) was adopted. S = 0.45,A C = 0.375) for graded verification. The results showed that 9 out of 10 samples were judged as "unqualified", of which 8 samples were due to safety score ψ S A score below 0.6 resulted in an immediate rejection, and one sample failed to meet the standard due to a comprehensive score below 60. The results indicate that the regulatory model, by strengthening the weighting of safety and consistency, significantly improved the sensitivity of identifying high-risk products, aligning with the core risk control needs of regulatory scenarios.

[0095] (2) Detection mode verification

[0096] Fifteen domestically produced general-purpose cosmetic products were selected and blind-tested by a third-party testing agency. The results showed that three products failed due to poor efficacy scores. E Two products were rated as "Basic Pass" due to their low scores, and two products were rated as "Integrity Pass" due to their low consistency scores. C Lower values ​​(pH fluctuations, differences in active ingredient content) were rated "Grade A". The testing agency reported that this model, by strengthening the weighting of experimental data (A... E + A C (Accounting for 60% in total), this significantly improves the screening efficiency for products with "exaggerated efficacy claims" and "unstable quality between batches," which aligns with the emphasis on objective data verification in testing scenarios.

[0097] (3) Validation of R&D model

[0098] In collaboration with three cosmetic companies, a research and development model (A) was adopted for prototyping 20 formulas under development. E = 0.325, A U The samples were graded based on a χ² = 0.26. The results showed that two samples were highly effective due to their high efficacy and positive user experience (ψ = 0.26). E ≥ 0.8, ψ U ≥ 0.85) triggered synergistic gain, resulting in a final score exceeding 95 points, and was rated "AAAAA level", highly consistent with the company's internal assessment; 3 other samples had lower experiential scores (ψ U A score < 0.6 indicates a failure to reach a higher level, suggesting the need to optimize skin feel or user experience. Company feedback shows that the grading results effectively differentiate the market potential of different formulas, providing a clear direction for product optimization. This validates the model's guiding value in the R&D process.

[0099] 4. Conclusion

[0100] This study constructs a cosmetic quality grading system based on four dimensions: safety, efficacy, quality consistency, and user experience. It proposes a hybrid model of "basic threshold + dynamic weight," achieving scientific grading through indicator standardization, dynamic weight matrix, and synergistic gain correction. This method exhibits high adaptability to various scenarios and good resistance to manipulation. Case studies demonstrate that the grading results highly align with market perceptions, proving the model's effectiveness and operability.

[0101] The detailed scoring and calculation methods for the four products in this invention are as follows: According to the grading index system of the present invention (Tables 1-4), the safety (ψ) of the four products S ), efficacy (ψ) E Consistency of quality (ψ) C User experience (ψ) U All scores are calculated through the process of "measured parameters → table lookup and scoring → weighted / arithmetic average".

[0102] I. Core Premise

[0103] None of the products have any "veto" items (such as the detection of prohibited substances or the exceeding of restricted substances). The secondary indicators for safety, quality consistency, and user experience are all calculated using an "equal-weighted average". The efficacy score was calculated as a weighted average of the claimed efficacy sub-scores, and all sub-scores were statistically significant (P<0.05).

[0104] Product A: Barrier Repair Moisturizing Cream (Skincare product, claiming "barrier repair + moisturizing")

[0105] 1. Security (ψ) S =0.95)

[0106] Score Calculation: The safety assessment comprises 8 secondary indicators, with a total score of 1.0 + 0.8 + 1.0 + 1.0 + 1.0 + 1.0 + 0.8 + 1.0 = 7.6. S =7.6÷8=0.95.

[0107] 2. Efficacy (ψ) E =0.90)

[0108] Score Calculation: Take the highest score for each efficacy (moisturizing 1.0, barrier repair 0.8), ψ E = (1.0+0.8) / 2=0.9.

[0109] 3. Quality consistency (ψ) C =0.90)

[0110] Score Calculation: There are a total of 6 secondary indicators, and the total score is 0.8 + 0.8 + 1.0 + 1.0 + 1.0 + 0.8 = 5.4, ψ C =5.4÷6≈0.90.

[0111] 4. User experience (ψ) U =0.83)

[0112] Score Calculation: There are a total of 7 secondary indicators, and the total score is 0.8 + 0.8 + 0.8 + 1.0 + 0.8 + 0.8 + 0.8 = 5.8, ψ U =5.8÷7≈0.83.

[0113] Product B: Hydrating Brightening Essence (Skincare product, claims to "remove blemishes and whiten skin")

[0114] 1. Safety (ψS=0.83)

[0115] Score Calculation: There are a total of 8 secondary indicators, and the total score is 1.0 + 0.8 + 0.8 + 0.7 + 1.0 + 0.8 + 0.8 + 0.7 = 6.6, ψ S =6.6÷8≈0.83.

[0116] 2. Efficacy (ψ) E =0.80)

[0117] Score Calculation: If there is only one efficacy claim, the highest score of 0.8 will be taken as the final score for that item. E =0.80.

[0118] 3. Quality consistency (ψ) C =0.70)

[0119] Score Calculation: The total score is 0.6 + 0.6 + 0.8 + 0.8 + 0.6 + 0.8 = 4.2, ψ C =4.2÷6=0.70.

[0120] 4. User experience (ψ) U=0.71)

[0121] Score Calculation: The total score is 0.8 + 0.6 + 0.6 + 0.8 + 0.6 + 0.6 + 0.6 = 5.0, ψ U =5.0÷7≈0.71.

[0122] IV. Product C: Firming and Anti-Wrinkle Face Cream (Skincare product, claims to be "anti-wrinkle")

[0123] 1. Security (ψ) S =0.95)

[0124] Score Calculation: Total score = 1.0 + 1.0 + 1.0 + 1.0 + 1.0 + 0.8 + 0.8 + 1.0 = 7.6, ψ S =7.6÷8=0.95.

[0125] 2. Efficacy (ψ) E =1.0)

[0126] Score Calculation: If there is only one efficacy claim, the highest score of 1.0 will be taken as the final score for that item. E =1.0.

[0127] 3. Quality consistency (ψ) C =0.97)

[0128] Score Calculation: Total score = 1.0 + 0.8 + 1.0 + 1.0 + 1.0 + 1.0 = 5.8, ψ C =5.8÷6≈0.97.

[0129] 4. User experience (ψ) U =0.94)

[0130] Score Calculation: Total score = 1.0 + 1.0 + 0.8 + 1.0 + 0.8 + 1.0 + 1.0 = 6.6, ψ U =6.6÷7≈0.94.

[0131] 5. Product D: Essential Oil Hair Shampoo (Shampoo and Conditioner, claiming to "condition hair + remove dandruff + prevent hair breakage")

[0132] 1. Security (ψ) S =0.78)

[0133] Score Calculation: Total score = 1.0 + 0.8 + 0.8 + 0.7 + 1.0 + 0.6 + 0.6 + 0.7 = 6.2, ψ S =6.2÷8≈0.78.

[0134] 2. Efficacy (ψ) E =0.77)

[0135] Score Calculation: The total score is 0.8 + 0.8 + 0.7 = 2.3, ψ E =2.3÷3=0.77.

[0136] 3. Quality consistency (ψ) C =0.70)

[0137] Score Calculation: The total score is 0.6 + 0.6 + 0.4 + 0.8 + 1.0 + 0.8 = 4.2, ψ C =4.2÷6=0.70.

[0138] 4. User experience (ψ) U =0.83)

[0139] Score Calculation: The total score = 0.8 + 1.0 + 0.8 + 0.8 + 0.8 + 0.8 + 0.8 = 5.8, ψ U =5.8÷7≈0.83.

[0140] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A cosmetic quality grading method based on dynamic weight optimization, characterized in that: Includes the following steps: Step 1: Testing and Result Scoring: Indicators for four dimensions of cosmetics—safety, efficacy, consistency, and user experience—are selected and tested to obtain measured values. Scores are assigned based on the test results, and the scores ψ for each of the four dimensions are calculated. S ψ E ψ C and ψ U ; Step 2, Weight Matrix Construction: By combining basic weight settings with pattern-based adjustments, the weight ratio of each dimension is allocated differently for different application scenarios such as supervision, testing, and R&D. At the same time, pattern enhancement coefficients are used to strengthen the core objective orientation and dynamically adjust the weights to form a weight method that takes into account both universality and specificity. Step 3: Calculate the overall score: Based on the scores of the four dimensions and the dynamically adjusted weights, calculate the overall score; Step 4: Cooperative gain correction: When both efficacy and user experience reach high standards, that is, ψ E ≥ 0.80 and ψ U When the score is ≥ 0.85, the product's final score is... λ is the gain score, calculated as follows: Step 5: Quality Grade Classification A comprehensive score greater than 95 points is AAAAA Excellent level; a comprehensive score of 85-95 points is AAAA Good level; a comprehensive score of 75-85 points is AAA Preferred level; a comprehensive score of 65-75 points is AA Standard level; a comprehensive score of 60-64 points is A Basic Pass level; a comprehensive score less than 60 points or in step one ψ S When the value is less than 0.6, it is considered as not meeting the standard.

2. The cosmetic quality grading method based on dynamic weight optimization according to claim 1, characterized in that: Step two is as follows: The basic weights for safety, efficacy, consistency, and user experience are 0.30, 0.25, 0.25, and 0.20, respectively. The basic weights are dynamically adjusted in different application scenarios: According to the core target of different application scenarios, the key dimension weight is improved, the scene exclusive enhancement coefficient β is introduced, the weight of the core dimension is amplified, the actual influence of the scene core dimension is strengthened, and the core dimension final application weight A x The calculation formula is as follows: in, β： A unique enhancement factor, ranging from 1.0 to 1.5; At the same time, the weights of non-core dimensions are proportionally reduced to ensure that the sum of the weights of all dimensions remains 1 after the adjustment. The specific formula is as follows: Where: A0: the original weight of the core dimension; A: The original weights of non-core dimensions; A X : application weight of core dimension; A X ′: application weight of non-core dimension.

3. The cosmetic quality grading method based on dynamic weight optimization according to claim 2, characterized in that: Under the regulatory model, the focus is on risk prevention and control. β The score is 1.5, with the core dimensions being safety and consistency. After calculation, the basic weights for safety, efficacy, consistency, and user experience are 0.45, 0.097, 0.375, and 0.078, respectively. In the testing mode, the focus is on objective data verification, with efficacy and consistency as the core dimensions. β The base weights for safety, efficacy, consistency, and user experience are 1.2, 0.24, 0.30, 0.30, and 0.16, respectively. Under the R&D model, innovation and market value are the guiding principles, with efficacy and user experience as the core dimensions. β The base weights for safety, efficacy, consistency, and user experience are 1.3, and after calculation, they are 0.226, 0.325, 0.189, and 0.26, respectively.

4. The cosmetic quality grading method based on dynamic weight optimization according to claim 3, characterized in that: The calculation method for the overall score in step three is as follows: Wherein: A S , A E , A C , A U is the application weight after dynamic adjustment; ψ S , ψ E , ψ C , ψ U is the score of each index.

5. The cosmetic quality grading method based on dynamic weight optimization according to claim 1, characterized in that: The safety indicators and their scores in Step One are as follows: Security score ψ S The method of calculation of ψ is as follows: Step 1: Check if there are any "veto" items; In the detection of banned substances, the control of restricted substances, the total number of colonies, mould and yeast, pathogenic bacteria, the evaluation of preservative efficacy, skin patch test and eye irritation test, if one of them scores 0, then S = 0, i.e. "one vote against", no need for subsequent calculation; Step 2: If there is no "veto" item, then take a weighted average of the scores of all secondary indicators to obtain ψ. S Score; If ψ after calculation S If the score is less than 0.6, it will be directly rated as substandard and no further evaluation will be conducted.

6. The cosmetic quality grading method based on dynamic weight optimization according to claim 1, characterized in that: The efficacy indicators and scoring rules are as follows: Efficacy score ψ E Calculation method: Based on the product's claimed efficacy, provide at least one type of valid evidence, either in humans or in vitro, and select 1-2 core indicators for scoring. If multiple types of evidence are provided, the highest score is taken as the single efficacy score. If the product claims multiple efficacy, the corresponding total score is calculated. This is the arithmetic mean of the scores of all claimed power components.

7. The cosmetic quality grading method based on dynamic weight optimization according to claim 1, characterized in that: The consistency indicators and scoring rules are as follows: Consistency score ψ C The calculation method is the arithmetic mean of each indicator with equal weight.

8. The cosmetic quality grading method based on dynamic weight optimization according to claim 1, characterized in that: The experiential metrics and scoring rules are as follows: User experience score ψ U The calculation method is: the arithmetic mean of each indicator with equal weight.

Images