Classification
Classification methods
- Manual classification by users
- Programmatic classification through APIs
- Template-based classification with entry templates
- Content-based automatic classification
- AI-powered classification through the Classification MCP server
Traditional classification
Entry templates help automate the classification process by filing documents into prespecified folders and predefining the class and property values of objects. This approach works well for structured workflows where classification rules are known in advance.
Content-based automatic classification examines document content to determine appropriate classification. The system supports automatic classification of XML documents through mapping scripts that associate XML tags with properties. Custom classification plug-ins can be created for other document formats.
AI-powered classification
The Classification MCP server provides AI-powered classification capabilities that go beyond traditional rule-based approaches. This server is part of the MCP (Model Context Protocol) layer in the Content Cortex architecture and enables intelligent content organization.
AI-powered classification capabilities include:
- Auto-classification
- Uses machine learning models to automatically classify documents based on content analysis. The system learns from existing classified documents to improve classification accuracy over time. This approach handles unstructured content more effectively than rule-based classification.
- Property extraction
- Automatically extracts relevant metadata from document content and populates object properties. The system identifies key information such as dates, names, amounts, and other business-relevant data without requiring predefined extraction rules.
- Metadata mapping
- Maps extracted information to appropriate property fields based on semantic understanding of the content. The system understands the relationships between different data elements and assigns them to the correct metadata fields.
AI-powered classification integrates with AI agents through the MCP protocol, allowing applications to request classification services programmatically. This integration enables intelligent document processing workflows that make classification decisions based on broader business context.
AI agent classification operations
AI agents in Content Cortex provide a conversational interface to classification capabilities. The Classification MCP server handles the backend intelligence (auto-classification, property extraction, metadata mapping), while AI agents make these capabilities accessible through natural language commands. Users interact with AI agents through the AI Agent plug-in, which coordinates with the Classification MCP server to process requests.
AI agents support the following classification operations:
- Suggest classification - Analyze document content and recommend the most appropriate document type. Users can ask the AI agent to examine a document and suggest which class it should belong to, ensuring correct classification without manual analysis.
- Reclassify document - Change a document's type when it has been miscategorized or its purpose has changed. Users can instruct the AI agent to reclassify documents through conversational commands, streamlining the correction of classification errors.
This conversational approach reduces classification time by up to 80% compared to manual methods. Users can classify documents accurately without understanding class hierarchies or property requirements, lowering training costs and accelerating adoption. Machine learning adapts to content patterns and improves accuracy over time, reducing rework and compliance risks. Organizations can scale classification operations across thousands of documents without proportional staff increases, and handle unstructured content that does not follow predefined templates, including legacy content repositories.