Three Oregon City calendars, reconciled into one review queue

1. Where the data comes from

Each publisher is read from its own public events endpoint, by a crawler that identifies itself and gives a contact address. Adding a publisher that runs the same calendar software is one line of configuration. Anything else needs its own adapter.

2. The categories problem, measured

Filtering by category only works if events carry categories.

3. Records a partner publishes with no match on the destination calendar

This is the comparison that is otherwise done by opening each partner site by hand.

4. Possible duplicates, surfaced rather than silently merged

The same event is often published under different names. Guessing wrong in either direction is costly, so similar records are linked and left for a person to resolve.

5. The reconciled set

Search and filter everything collected. Each card shows which publishers it came from, so no record is unattributed.