During Data Loss Prevention detection processes metadata (file properties such as title, subject, author, etc) is not detected and reported on after the server is initially setup causing some keywords to not get detected.
By default metadata is not detected. This option must be turned on in Advanced Server Settings.
Set the "ContentExtraction.EnableMetaData" value to "on", without quotes, to detect metadata from Microsoft Office and PDF files. Extensible Metadata Platform (XMP) content is not detected. Note that enabling this metadata detection option can cause false positives.