Metadata - Label/Tag Definitions
GEODI Classifier has a flexible labeling mechanism to adapt a DLP software or an existing labelling schema.
PDF, Office, Libre Office documents and OWA mails are labeled
For other content types, if NTFS is used, ADS is used;
Linux, document properties are used.
For DBs a value and defined column is used for labelling.
GEODI directly supports files tagged with your current classification tool and adapts to the existing label scheme.
Label Definitions
Classes are created as described on the following page:
Class Definitions
Each class requires a label definition. It is recommended to use the standard class definitions provided, which you can modify later. Definitions and sample files are available there. In this case, you do not need to define labels manually. The provided default classes explained at Metadata - Label/Tag Definitions
If you have a legacy classifier, the best practice is to provide a Word document for each class and define labels accordingly.
Labels must follow the name:value format. Each name must be unique, and the first value for each class must also be unique, as it defines the class itself.
name1:value1
name2:value2
...The value can take many different forms. You can use variables inside the labels. Variables
class:confidential
class:{58f30e89-66db-4092-a81f-282a2eee431c}
class:<x class="confidential"></x>
class:{"id":"58f30e89-66db-4092-a81f-282a2eee431c"}
class:<x>confidential</x>
user:%User%
userandtime:%User%-%Time%
These labels are the same along Office, PDF, Libre Office files, or ADS.
ADS and Custom Attributes
For file types that don't support tags, tags are stored in a platform-dependent manner.
On Windows with NTFS, Alternate Data Streams (ADS) are used; on macOS, Linux, and FreeBSD, the Custom Attribute feature is utilized.
You can query these tags using built‑in operating system commands:
Windows (ADS) →
dir /rmacOS →
xattr -l <file path>Linux →
getfattr -d <file path>FreeBSD →
listextattr user <file path>
The portability of ADS and other tags is limited, and DLP systems must also support them.
Default Classes
If you do not have an existing labelling schema, the following classes and tag definitions come standard with GEODI. You may directly use or modify the rules. If there is no existing tag, you can use the tags below for DLP compliance.
Class | Labels | Coverage |
|---|---|---|
Confidential |
| Has money and money greater than 50K USD or equivalent and selected keywords. Selected Keywords are in a Disctionary. You can freely modify the disctionary, named |
PII |
| (A name or ID) and (Phone, E-mail, Adress, or Bloodtype) ID covers National ID numbers, Medical Numbers, Passport, and the like. |
Restricted |
| IBAN, Tax Numbers, SWIFT code and keywords like “restricted”, “internal use only” Keywords are in a dictionary. You can freely modify the dictionary, named |
Unclassifed |
|
|
**