Metadata Anomalies

From UA Libraries Digital Services Planning and Documentation
Jump to: navigation, search

Contents

Conflicts between contents of box and metadata

Item not in metadata: Manuscripts/auto-numbered by us

  • First, do a "find" search in the Title column of the spreadsheet, looking for dates or names that appear in the item; this is to make sure that it's not already in the metadata and simply physically filed in the wrong place
  • If you don't find any metadata for that item, capture it anyway, assigning it a number at the end of the file naming sequence for the collection
    • check the metadata to confirm what that number would be
    • check spreadsheets and web directory to make sure it hasn't been used
  • Create a line of metadata for that item at the end of the spreadsheet, using the newly assigned number; it's okay to integrate it with the current batch

Item not in metadata: Photos/pre-numbered

  • This item has been culled from the spreadsheet by April or someone else before it gets to us. Check to see if we're digitizing skipped items or not.
  • If you do capture the item, insert the missing number into the metadata spreadsheet and do your best to extrapolate metadata from nearby lines.

Item not in box

Make a note in the log data. Obviously, you can't capture what's not there.

Conflicts between item and metadata

Metadata inconsistent

  • Examples: name is different in Title and Creator; date is different in Title and Date
  • Fix the incorrect field, based on examination of the item
  • Make a note of what you fixed in the Metadata Changed column in the log data

Item has fewer/more pages than metadata indicates

In general, page numbers in the Format column are not necessarily going to be correct when they come to us, since the archivists don't always know what we'll consider a page; for books, page numbers are usually estimates to begin with. Expect to amend the metadata frequently. However...

  • If there are way fewer pages than metadata
    • Check to see if anything is loose in the folder.
    • If you can't determine where the missing page might be, scan what you do have, amend the page count in the Format column, and add a note to the Description column: "Item had pages missing."
  • If there are way pages than metadata
    • Check object against metadata to make sure the page actually belongs to that item.
    • If it belongs, amend the page count in the Format column.

Date in metadata does not actually exist

  • Check to see if it's a data entry typo by the archivists or an error by the creator
  • If typo, fix it
  • If error, leave it the way it is in the Title and Date fields, but add a note to the Description field: "Date as given in item"
    • This policy is per Donnelly (email to Kate, 9/5/13): metadata should reflect what is written (maybe with a note that says essentially "yeah, we know there’s no Feb. 30th").

Metadata Incorrect

  • If you see a typo or otherwise obvious error, fix it.
  • If you see a misspelling, make sure it's actually a misspelling and not just an oddity.
  • If you think something looks normal but wrong, do some research or get a second opinion. You might be misreading the item or misunderstanding how the information is supposed to be recorded for that kind of item. When in doubt, contact the person who wrote the metadata or another archivist.
  • Examples:
    • Date says April 5 but the receipt says April 1. Look more closely -- sometimes receipts have two dates, and if April 5 is in there somewhere, there's a reason it was recorded.
    • Letter addressed to Sister, but you know good and well who the sister is, based on other items. Don't correct this -- sometimes the archivists will refrain from guessing when identifying people, or they'll use the actual salutation in the Title field but a real name in the Recipient field.
    • City is given as Tuskaloosa, not Tuscaloosa. Maybe correct this, depending on what field it's in -- that was the old spelling, so it's okay to use it in a title or uncontrolled field, but you'll want the Location to have the correct modern spelling.
    • Date for a photograph is 1679. Correct this -- fill in the real date if you can reasonably work out what it is (if similar photos nearby are 1979, for example), but if you can't, leave it blank or give an approximation (if you think it could be 1979 or 1967, for example, give both decades, 1960-1979).
    • Item gives county name but not the state it's in. Research this -- you may think you know where it is, but there may be more than one in the country and it's unclear which this is, in which case you'd not want to guess.
Personal tools