manufacturing-failure-reason-codebook-normalization

To normalize manufacturing defect failure reasons, this skill segments raw text, scores codebook candidates using station compatibility and context alignment, and outputs a single standardized code with calibrated confidence per segment.

Codebook candidate scoring evaluates segmented failure text by measuring station compatibility and context alignment against the codebook, applying deterministic tie-breaking to select the best matching code with a calibrated confidence score.

Standardizing inconsistent failure reason text uses station-aware validation during semantic matching to ensure segmented defect reports conform to product codebooks regardless of which station recorded the original failure.

Yes, you can batch process raw failure reasons by providing multiple records, and the skill will segment text, score codebook candidates, and output standardized codes with explanations and confidence scores for each segment.

Confidence calibration assigns a score to each predicted code by evaluating evidence such as station compatibility and context alignment, producing auditable confidence levels for every standardized defect code output.

Automated defect reason normalization depends on the quality of the provided product codebooks and raw failure text; ambiguous segments may yield lower confidence scores, requiring manual review for accurate codebook alignment.