In order for products to be classified semi-automatically, training sets [Training-Sets] must initially be stored for classes, on the basis of which a geometric comparison is then carried out.
The following diagram shows an overview of the process:
![[Note]](https://webapi.partcommunity.com/service/help/latest/pages/jp/partsolutions_user/doc/images/note.png) | Note |
|---|
Semi-automatic classification and Classwizard have the same aim, but the approach and target group are slightly different.
Classification Wizard (call in PARTdataManager or CAD ) Moving or inserting individual products / product families into an existing structure Class suggestions based on existing products in the target catalog Assignment of class attributes during execution
-> More user input required -> Understanding of the structure required
Semi-automatic classification (GUI)
Moving many products into existing structure Class suggestions based on selected training sets (structure can be empty; training set can come from another catalog) Automatic shifting for best results (> minimum similarity) possible (theoretically for any number of parts simultaneously)
-> One-time creation of training data required by advanced user -> Execution can be carried out by anyone without a deeper understanding of the structure (apart from confirming ambiguous assignments)
Semi-automatic sizing (pipeline)
Inserting/moving many products into existing structure (both with ready-made training sets and using the existing products in the structure) Automatic move/insert for best results (> minimum similarity) possible (theoretically for any number of parts simultaneously) Copying of existing structure during the creation of a new structure (original filling)
-> Can be used for rebuilding the structure (e.g. based on standard catalog) -> No user input required for updates (e.g. when parts are updated 24 hours a day, they are automatically moved to the appropriate classes if suitable ones are found)
|
This chapter describes semi-automatic sizing.
