Visual Review Feature Extraction and Defect Triage Matrix

Role: Principal Visual Merchandising Analyst with expertise in computer vision auditing, customer experience feedback loops, and automated catalog enrichment.

Context

• Product Line: {{product_line}}
• Visual Attributes Checklist: {{visual_attributes_list}}
• Customer Sentiment Dataset: {{customer_sentiment_dataset}}
• Quality Defect Criteria: {{defect_detection_criteria}}
• Multimodal Model Engine: {{multimodal_model_engine}}
• Merchandising Display Priority: {{display_priority}}

Task

Develop a granular visual feature verification matrix that maps user-generated review photos against textual feedback and product specifications to quantify defect signals and prioritize high-converting customer imagery.

Method

1. Review {{product_line}} catalog specifications to define ground-truth visual benchmarks.
2. Dissect {{visual_attributes_list}} to establish observable image tokens (color match, seams, finish, sizing).
3. Cross-reference visual findings against sentiment trends provided in {{customer_sentiment_dataset}}.
4. Apply {{defect_detection_criteria}} to categorize visible product failures (damage during shipping, manufacturing flaws, wear-and-tear).
5. Score visual UGC assets for front-end merchandising suitability based on {{display_priority}}.
6. Configure multimodal prompt parameters for {{multimodal_model_engine}} to automate image attribute extraction.
7. Synthesize findings into a cross-functional triage matrix aligning QA, merchandising, and customer support.

Constraints

• Visual defect classifications MUST be tied to verifiable, observable physical indicators.
• Assets marked for promotional display MUST meet both high resolution and positive sentiment thresholds.
• Do not infer sentiment without direct visual or textual evidence in the provided data.
• Every matrix row MUST state a specific downstream action for Product Engineering or Merchandising.

Output format

• Analytical Overview (Concise summary of visual-to-sentiment alignment)
• Feature & Defect Triage Matrix (Markdown table with 7 columns: Feature/Defect Category, Visual Clue Description, Text Sentiment Correlation, Severity/Value Score, Recommended Tag, Routing Department, Merchandising Status)
• Multimodal Inference Template (Structured JSON/Prompt schema for automated pipeline execution)

Self-review

• Ensure all variables ({{product_line}}, {{visual_attributes_list}}, {{customer_sentiment_dataset}}, {{defect_detection_criteria}}, {{multimodal_model_engine}}, {{display_priority}}) are represented in the analysis logic.
• Verify that positive merchandising tags are clearly separated from product quality defect flags.
• Check that the triage recommendations provide clear, deterministic outcomes for catalog teams.