Samples of information assets (documents, files, etc.) that have their fingerprints collected and stored in the database are referred to as fingerprint sources. Samples may be modified, added, or removed, or their secrecy level may change with time. For the database to accommodate all those changes, the server runs classification tasks on a regular basis, which will update the fingerprints storage as described below.

Information samples are processed and their fingerprints are created by tasks on the DeviceLock Enterprise Server. Each task attributes to a certain classification, and assigns it to the fingerprints that it creates. For instance, fingerprints created by a task of “Confidential” classification are attributed to that same “Confidential” classification.

On every run, the task may inspect files in a particular folder. For each file, the task first creates the fingerprints of the file and compares them with the fingerprints from the database. The further processing of the file’s fingerprints depends upon comparison results such as:

Even if the source of a fingerprint is removed, the fingerprint remains in the database. DeviceLock administrators can delete fingerprints or their individual versions by hand using the DeviceLock Management Console.

The versioning threshold determines whether to create a new fingerprint or merely add a new version to an existing fingerprint. The DeviceLock Enterprise Server specifies separate thresholds for text content (such as text files) and for binary content (such as image files).

Many files hold content of both types. For instance, Microsoft Word documents are binary files that can contain text and images. Fingerprints of files with mixed contents hold elements identifying text content and elements identifying binary content. When classifying such a “mixed” fingerprint, the server applies both thresholds, and separately assesses the match percentages for “text” and “binary” elements of the fingerprint. This results in the following effects: