COBBLESTONE SOFTWARE
Contract Insight — User Guide
Category Training Data
03. Admin User Manual › VISDOM & Risk Management › Category Training Data
Note: Each procedure in this guide begins at the Contract Insight homepage, so any section can be followed on its own. Custom fields are configurable, so the exact field list on a contract category record may vary between environments.
1. Overview
VISDOM's text recognition is tuned by reference lists maintained in Master Reference Data rather than by retraining a model from scratch. Three tabs there hold this training data: VISDOM Business Name Words, VISDOM Noise Words, and VISDOM Regular Expressions. Business Name Words is a maintained list of terms (such as "Inc.", "LLC", or industry-specific suffixes) that help VISDOM recognize company names inside contract text; Noise Words lists terms to disregard when scanning for meaningful content; VISDOM Regular Expressions stores pattern-matching expressions administrators can add to sharpen recognition of structured values (dates, IDs, dollar amounts, and similar patterns) specific to an organization's documents. Together these lists let administrators tune what VISDOM notices without touching the underlying AI model configuration.
2. Key Elements
- Master Reference Data > VISDOM Business Name Words — terms that help VISDOM recognize organization/company names in text.
- Master Reference Data > VISDOM Noise Words — terms VISDOM should disregard when scanning documents.
- Master Reference Data > VISDOM Regular Expressions — custom pattern-matching rules for structured values.
3. Quick Reference Summary
[TODO: content needed]