Evaluation Methodology & Scoring System

1. Executive Summary & Guiding Principles

Cosmetiqly operates as an independent, computational cosmetic safety and dermatological compliance database. In modern retail cosmetics, consumers are routinely exposed to subjective marketing buzzwords such as “clean,” “pure,” “hypoallergenic,” and “dermatologist-approved.” In the vast majority of international jurisdictions, including the United States Food and Drug Administration (FDA) and the European Commission, several of these terms lack statutory legal definitions. As a direct consequence, consumers managing acute dermatological conditions—such as Malassezia folliculitis (fungal acne), contact allergic dermatitis, barrier impairment, rosacea, or gestation-specific chemical exclusions—are left without a deterministic, objective standard for evaluating physical formulations.

Cosmetiqly was engineered to resolve this informational asymmetry by replacing marketing conjecture with computational taxonomy. Our platform adheres strictly to four core technical tenets:

• Authentic Formulation Data Only: We never generate, extrapolate, or estimate product ingredient lists. Every single product record in our repository is derived from verified physical packaging barcodes cataloged within open, verifiable databases (Open Beauty Facts under the Open Database Licence). If verified packaging data does not exist for a product, that page does not exist on our site.
• Deterministic Exact Token Matching: We strictly prohibit loose substring matching algorithms. Substring matching is the primary technical flaw responsible for high false-positive rates across legacy cosmetic rating portals.
• Universal Mathematical Scoring: Every numeric score from 1.0 to 10.0 is calculated via a single, transparent, published mathematical formula. No commercial entity or cosmetic manufacturer can purchase score adjustments, sponsor placement, or alter verdict determinations.
• Peer-Reviewed Literature Citations: Every flagged substance is explicitly linked to published monographs from institutional safety bodies—such as the Cosmetic Ingredient Review (CIR), the European Commission Scientific Committee on Consumer Safety (SCCS), the FDA, and indexed dermatological journals.

2. Rule Engine v2 & Exact Token Normalization

The primary diagnostic engine driving Cosmetiqly is Rule Engine v2. When a cosmetic formulation is ingested into our processing pipeline, its International Nomenclature of Cosmetic Ingredients (INCI) text undergoes comprehensive lexical and structural normalization prior to evaluation.

Lexical Parsing Architecture

Cosmetic packaging presents unique typographical challenges, featuring multilingual translations, botanical binomial nomenclature, trade names, and varying punctuation formats. To ensure uniform matching against toxicological reference databases, our normalization pipeline executes the following deterministic phases:

1. Parenthetical and Bracketing Extraction: Cosmetic manufacturers frequently nest commercial marketing names or translations inside parentheses (e.g., Water (Aqua/Eau), Glycerin (Vegetable Derived), or Tocopherol [Vitamin E]). Our tokenizer strips non-standard modifiers while preserving the core standardized chemical designation (water, glycerin, tocopherol).
2. Standardized Delimiter Splitting: In compliance with global labelling directives, ingredients are delimited by commas, semicolons, bullets, or periods. The parser splits raw string sequences into discrete atomic token entities.
3. Punctuation and Numeric Sanitization: Incidental punctuation, asterisk references denoting organic certification (e.g., *Certified Organic), and trailing percentage signs are systematically removed.
4. Case Invariance and Whitespace Collapse: All tokens are converted to lower-case UTF-8 and collapsed to single-space word boundaries to eliminate whitespace variance across international packaging variations.

Elimination of the Substring Fallacy

A pervasive structural bug in first-generation cosmetic directories was the reliance on substring inclusion methods (e.g. JavaScript's string.includes()). Under a substring-based paradigm, searching for the trigger alcohol inadvertently matches cetearyl alcohol, stearyl alcohol, and behenyl alcohol. In reality, while simple denatured alcohol (alcohol denat.) is a volatile solvent known to disrupt stratum corneum lipid bilayers, cetearyl alcohol is a benign, emollient fatty alcohol that actively supports skin hydration.

Rule Engine v2 completely eliminates this bug by mandating strict token equality: an ingredient is flagged if and only if its fully normalized INCI token matches the exact chemical trigger token in our verified blacklist.

4. Binary Verdict Framework: Free-From vs. Risk-Based Rules

Cosmetiqly organizes its 15 safety criteria into two fundamental functional paradigms:

Category A: “Free-From” Binary Exclusion Rules

Rules such as Silicone-Free, Fragrance-Free, Paraben-Free, Sulfate-Free, Vegan & Cruelty-Free, Reef-Safe, Gluten-Free, and Phthalate-Free represent definitive compositional boundaries. In these categories, the presence of even a single trigger represents a definitive failure of the standard:

• Trigger Count = 0: Verdict = PASS (“Free of flagged ingredients”).
• Trigger Count ≥ 1: Verdict = FAIL (“Fails standards”).

For example, a product containing parfum or linalool fails the Fragrance-Free rule unconditionally, regardless of its score on other criteria.

Category B: “Risk-Based” Clinical Rules

Standards such as Fungal Acne (Malassezia), Pregnancy & Lactation, Non-Comedogenic, and Rosacea & Eczema evaluate multifactorial biological thresholds where ingredient severity varies:

• High-Risk Triggers ≥ 1: Verdict = FAIL (“Fails safety standards”). The formulation contains active substrates or irritants directly documented to induce pathology.
• Moderate Triggers ≥ 1 and High-Risk = 0: Verdict = PASS WITH CAUTION (“Review the flags”). The formulation lacks direct primary triggers but includes potential cofactors or concentration-dependent agents.
• All Triggers = 0: Verdict = PASS (“Pass”).

5. Category Relevance Matrix

A common failure of low-quality programmatic SEO directories is cross-multiplying every product by every rule, generating absurd queries such as “Is this lipstick reef-safe?” or “Is this hair conditioner pregnancy-safe?”

Cosmetiqly implements a strict Category Relevance Matrix that restricts rule evaluation to biologically and physically meaningful domains:

If a product category is not dermatologically relevant to a rule, the combination is rejected at the routing layer and returns a true RFC 404 response.

6. Scientific Literature Hierarchy & Citation Standards

Every trigger in our database must possess an authoritative, traceable peer-reviewed or regulatory reference. We structure our evidence hierarchy into four institutional tiers:

1. Tier 1 — Regulatory Statutes & Legal Directives: European Commission Cosmetics Regulation (EC) No 1223/2009 Annexes II, III, and IV; Hawaii Senate Bill 2571 / Act 104; US FDA Monograph Regulations.
2. Tier 2 — Institutional Safety Opinions: Cosmetic Ingredient Review (CIR) Expert Panel Assessments; EU Scientific Committee on Consumer Safety (SCCS) Final Opinions.
3. Tier 3 — Clinical Dermatology Literature: Peer-reviewed prospective and retrospective studies published in leading indexed journals, including the Journal of the American Academy of Dermatology (JAAD), British Journal of Dermatology (BJD), and Contact Dermatitis.
4. Tier 4 — Specialist Medical Societies: American Academy of Dermatology (AAD), American College of Obstetricians and Gynecologists (ACOG), National Rosacea Society (NRS), and American Contact Dermatitis Society (ACDS).