Mainly which are the most active users that upload/search documents and which are their most repeated metadata values in their search and uploads. This way a more detailed classification values can be given to them so they can navigate the documents with more precision.
I.e. there are 200 documents with the same metadata values that are frequently readed/updated , that may be because it's metadata field is to generic so we try to find more specific atributes of this documents . Then 200 other documents have also a same metadata but searches that found them are few and usually succesful so we leave it's metadata values as they are. I would count how many searches leaded to a re-search (search->search->search->visualization/download) or a big amount of results (search->scroll->scroll->scroll->visualization/download) Also it would be interesting to track which users are uploading many documents with the same metadata fields so someone can check if they are doing it well or if they are just beeing lazy , specially when users with the same role are classifing in a more detailed way. This can be complicated a lot so I would look for the simplest solution as a first approach: track the size of search results along with the users who made it and check if there's any correlation between similar users or with the users past searches. I.e users in the account department allways search fore receipts of a certain kind of product ( paper ) that result into lot's of receipts . Noticing this a new metadata can be added specifying the product kind ( white paper , brown paper ) ... Hope I explained it well, and sorry for my english --
