Classification systems label documents by category based on content. Legal applications include document type identification, issue tagging, responsiveness determination, and privilege screening.
Systems may use supervised learning from labeled examples or apply a language model directly.
Alternative Names:
Document Categorization, Auto-Classification
Why it Matters?
Classification is the workhorse application in document-heavy litigation, and the accuracy threshold depends entirely on the consequence of error. Sorting documents into type categories for organization tolerates mistakes that responsiveness or privilege determination does not. Matching the required accuracy to the task, rather than accepting a single system-wide performance figure, is what makes deployment defensible.
Frequently Confused with
Related terms
Frequently asked questions
What accuracy is required for classification?
How does this differ from technology-assisted review?
