Creating an Analytics Categorization Set
Relativity Analytics Categorization Sets are sets of example documents that can be used as a basis to group together conceptually similar documents. The same document can be placed in up to five different categories. Categorization Sets are needed for Assisted Review (Relativity's term for TAR). Categorization will assign a rank to each document with respect to how similar it is to the examples in the categorization set.
The documents in a set should ideally represent a single concept and should possess at least one or two full paragraphs on the concept. (Examples can be selections of text from a document.) Documents with headers, footers, or poor OCR should be excluded. No more than 50,000 example documents should be used in a categorization set.
After creating a saved search with the documents that are to be categorized, follow these steps:
1. Under Indexing & Analytics, select Analytics Categorization Set and then click 'New Analytics Categorization Set'.
2. Name the set, select the saved search, and designate an Analytics Index.
3. Choose a minimum coherence score - this determines how similar a document must be to the example documents to be categorized.
4. Maximum categories per document - If a value higher than 1 is entered you will not be able to sort on the Category Rank field
5. Categories and Example Source - this should be a multiple or single choice field. Any choices associated with this field will automatically be added as categories. Any documents where the field is set will be added as example documents. The 'Auto Synchronize on Categorize All' must be set to 'Yes' for this to work.
6. Example Indicator Field - this should be a Yes/No field. Any documents which have 'Yes' selected will be used as example documents.
7. After saving the categorization set, add an analytics category . . .
8. . . . and an example document (or example text) for that category. (Any text that is entered will override a selected document.)
9. You can also add an example document directly from the Viewer.
10. After example documents have been added, the 'Categorize All Documents' button on the console will be enabled. (The 'Create Categories and Examples' button on the console will be enabled if example documents have not been manually added, and you're using the 'Categories and Example Source' option instead.)
Click on 'Refresh Page' to check on the status of categorization. The page may not automatically update.
11. After the categorization is completed, two new fields named, 'Categories' and 'Category Rank' will be created