For each product that arrives at the node's input, the node proceeds as follows:
The PARTsolutions catalog is searched using the selected geometric search template to find similar parts .
Based on the similar parts, the node determines the best-fitting class. It is not just the most similar match that counts: A class with many very similar parts can outrank a class with a single, somewhat more similar part.
If a usable fingerprint is missing or if no sufficiently similar class is found, the specified fallback directory is used instead.
The product is assigned a classification with the system ID
pdm; if it already exists, it is reused. Two attributes are set in it:
The products supplemented in this way are passed on to the next node via the output. Other documents in the input data remain unchanged.