Refreshing a crawl attempts to find new data, modified data, and data that has been removed without doing all of the work required of a completely new crawl. Specifics of what may be refreshed depends somewhat on the connector (see Connector and Seed Features). To refresh a crawl, you must have a live sub-collection. Choosing the refresh action in the Live Status heading will cause the refresh to occur in place in the live sub-collection. Choosing refresh in staging will copy the live data to staging and then perform the refresh in the staging sub-collection.
Refreshing a crawl does not happen automatically: to refresh a crawl, you must either schedule automatic recrawls or manually initiate one by clicking a refresh button on your search collection's Overview tab in the Watson Explorer Engine administration tool.