CONCEPT-BASED CATEGORIZATION

How It Works

The conceptual searching power of CAAT can be harnessed to rapidly classify or organize documents based on topics or subject matter, in a human-assisted process called Concept-based Categorization.

Concept-based Categorization is designed for users who already know what key issues or topics they need to find. The mechanism behind Concept-based Categorization is both elegant in design and simple in deployment: CAAT uses user-defined examples of issues or topics and rapidly compares new documents to these examples to logically classify those documents.

Users who already know the key issues or topics generally have one or more documents that exemplify that issue or topic. These documents become examples, and the topics become the categories. Only a handful of relevant examples need to represent a particular category; in fact, the purpose of these examples is to “describe” what ought to be placed in that category. Many users don’t even include full documents (they edit out extraneous content so the examples are especially concise).

Concept-based Categorization is exceptionally fast and requires a very small footprint, since a search index is never created. It can churn through millions of documents at breakneck speed and quickly sort those documents into user-defined categories for further processing and review.

How it is Used in eDiscovery

Very early on in any case, both sides know the key topics and issues. But what they don’t know are which documents in ESI being processed pertain to which topics and issues. Concept-based Categorization or Classification addresses that by extending what these knowledge workers already know about the case across all the documents being presented as “possibly responsive.”

Concept-based Categorization’s flexibility also benefits eDiscovery. Documents that are relevant to multiple topics, concepts, or issues can be categorized accordingly (i.e., in multiple “buckets”) and a relevance score instantly tells users just how relevant that document may be be. CAAT’s categorization also provides numerous thresholds, meaning that documents which aren’t particularly relevant to any user-defined category can be placed in an “uncategorized” bucket for later review or discard.

CAAT’s concept-based categorization operates at a mathematic level: it is repeatable, easily fine-tuned, and unlike human-based classifiers or coders, CAAT never gets tired, forgetful, or judgmental.