Confidence

Confidence expresses how certain the API is about a piece of extracted information. The model wraps a single numeric score between 0.0 and 1.0. On dimensions, bores, chamfers, threads, radii, roughness and GD&T frames it is 1.0 when nothing about the reading is in doubt, 0.8 or above for a reasonable reading, and below 0.8 when the API found a concrete reason to doubt it. Results at 0.8 or above can be consumed automatically, while lower scores should trigger human review or cross-checks against ERP data. See Understanding Confidence Scores for what lowers a score, and for the objects whose scores do not follow this scale yet. By exposing this signal, the API enables transparent decision-making and traceability, which are essential when automating workflows that historically required manual validation. Treat the Confidence class as a signal for prioritizing work queues, designing escalation rules, or measuring how changes in drawing templates affect parsing accuracy over time.

63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81

class Confidence(BaseModel):
    """
    Represents a confidence score for a feature or measurement.
    """

    score: Decimal = Field(
        ...,
        description=(
            "The confidence score of the reading, from 0.0 to 1.0. On "
            "dimensions, bores, chamfers, threads, radii, roughness and GD&T "
            "frames, a reading with nothing to doubt scores 1.0, and 0.8 or "
            "above is a reasonable reading. Below 0.8, at least one concrete "
            "reason to doubt it was found, such as a tolerance that is on the "
            "drawing but could not be read, or a value that no standard series "
            "contains, and the reading is worth checking. Scores on other "
            "objects, such as notes, do not follow this scale yet."
        ),
        examples=[Decimal("0.95")],
    )