Collect
Official catalogs, product pages, drawings, and manufacturer documentation are registered as immutable source records.
The RCPS approach
Comparable data
without flattened meaning.
Engineering parameters are useful only when their units, operating conditions, product scope, and evidence survive normalization.
Our principle
Normalize the value.
Preserve the context.
“40 Nm” is not a complete fact until you know whether it is rated, peak, emergency, or average-load torque—and under which speed and duty conditions.
Data pipeline
From source document to public record.
Observe
Every extracted value remains attached to its raw name, unit, page location, and capture context.
Normalize
Values map into a category-specific parameter dictionary with standard units and explicit semantic boundaries.
Validate
Schema checks, category rules, cross-record constraints, and human review make failures visible.
Publish
Only approved fields enter the public projection, with quality status and source summaries retained.
Publication boundary
Unknown is better than wrong.
Missing, conflicting, or semantically ambiguous values are held back or shown explicitly as unavailable. Empty data never becomes a silent success.
- Strict public-field allowlist
- Unit and type validation
- Source and capture timestamps
- Explicit provisional status