Administration Guide › Content Registration › How to Set Up Content Agents › To create a content agent › What Level of Accuracy Do I Need?

What Level of Accuracy Do I Need?

(Applies to Text Detection content agents only)

The accuracy of an agent depends on the level of detail in the document fingerpints. If you include more detail in a document's fingerprint, the agent is more accurate. That is, the agent can recognize content originating from that document more reliably.

At one extreme, you can generate a fingerprint that covers every sentence in a protected document. Agents based on highly detailed document fingerprints like these are the most accurate.

At the other extreme, you can generate a fingerprint that slices the document into sections analgous to a page. This fingerprint only covers the most significant phrases within each page-sized section. Agents based on these fingerprints are less accurate, but have far smaller indexes.

Note: A content index containing highly detailed (sentence-level) fingerprints can be very large indeed. In fact, these indexes can be eight times larger than comparable indexes containing the least detailed (page-level) fingerprints. Such large indexes can be a particular problem if you enable content agent triggers on your CA DataMinder endpoint machines, because each endpoint machine maintains a copy of each agent's content index.