As part of the score boosting work for #72 it has become necessary to track which source object type a document was obtained from. Without this we don't know which object type's scoring metadata to update when removing an item.
Proposal:
When using WithObjectTokenization, track an internal unique object type id, similar to how we uniquely track field ids.
Update ItemMetadata (Soon to be DocumentMetadata #92) to track the object type id it was stored for. For loose text documents, no object type id will be stored.
Persist this information with the serialized index.
As part of the score boosting work for #72 it has become necessary to track which source object type a document was obtained from. Without this we don't know which object type's scoring metadata to update when removing an item.
Proposal:
WithObjectTokenization
, track an internal unique object type id, similar to how we uniquely track field ids.ItemMetadata
(Soon to be DocumentMetadata #92) to track the object type id it was stored for. For loose text documents, no object type id will be stored.