> ## Documentation Index
> Fetch the complete documentation index at: https://docs.graphora.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Complete Platform User Guide

> Step-by-step guide for new users to get started with Graphora UI Frontend

# Complete Platform User Guide

This comprehensive guide will walk you through the entire Graphora platform workflow, from initial setup to successfully managing your knowledge graphs. Follow these steps in order for the best experience.

## Step 1: Configure Database and API Settings

When you first access Graphora, you'll need to provide your database details and AI API key on the **Config** page.

### Required Configuration

1. **Navigate to the Config page** in your Graphora dashboard

2. **Database Configuration**: Provide your Neo4j database connection details:
   * **URI**: Your Neo4j database URI (e.g., `bolt://localhost:7687`).
   * **Username**: Database username
   * **Password**: Database password
   * **Environment**: Choose between staging and production databases

3. **AI API Key**: Configure your Gemini API key for AI-powered processing:
   * **Gemini API Key**: Your Google AI Studio API key
   * This enables AI-powered document processing and knowledge extraction

<Note>
  Graphora follows a BYOK (Bring Your Own Key) architecture, meaning your API keys and database credentials remain under your control and are never stored on Graphora's servers.
</Note>

### Validation

After entering your configuration:

* Test your database connection to ensure it's working
* Verify your Gemini API key is valid
* Save your configuration before proceeding

***

## Step 2: Create or Import Your Schema

Next, you'll need to define the structure of your knowledge graph using the **AI Copilot** page.

### For New Users (No Existing Schema)

Use the [Schema Copilot](/platform/schema-copilot) -- a chat-based AI assistant that builds ontology schemas from natural language descriptions.

1. **Navigate to the AI Copilot page**
2. **Describe your domain**: Provide a clear description of what you want to model
   * Example: "I want to model a company's organizational structure with employees, departments, and projects"
3. **Use AI to generate schema**: The copilot uses streaming responses to build your schema in real time
4. **Refine through conversation**: Ask follow-up questions to adjust entities, properties, and relationships
5. **Save your ontology**: The system will validate and store your schema

### For Users with Existing Schemas

If you already have an ontology definition, it must adhere to the **Ontology Editor YAML format**:

<Card title="Ontology YAML Format" icon="file-code" href="/client/ontology-editor/yaml-syntax">
  Learn about the complete YAML syntax, structure, and requirements for ontology definitions.
</Card>

**Key Requirements**:

* Use **camelCase** for all property names
* Follow the version 1 format specification
* Include proper entity definitions with properties and relationships
* Define quality rules for data validation

**Example Structure**:

```yaml theme={null}
version: 1
entities:
  Company:
    properties:
      companyName:
        type: string
        description: Official name of the company
        required: true
        unique: true
      foundedYear:
        type: integer
        description: Year the company was founded
    relationships:
      HAS_EMPLOYEE:
        target: Person
        cardinality: one_to_many
```

***

## Step 3: Transform Documents into Knowledge Graphs

Once your schema is ready, move to the **Transform** page to process your documents.

### Document Upload and Processing

1. **Select your ontology**: Choose the schema you created or imported
2. **Upload documents**: Support for multiple formats:
   * PDF files
   * Text files (.txt)
   * Markdown files (.md)
3. **Start transformation**: The system will:
   * Extract text from your documents
   * Use AI to identify entities and relationships
   * Structure the data according to your ontology
   * Generate a knowledge graph

### Monitoring Progress

* **Real-time status updates**: Track the transformation progress
* **Resource metrics**: Monitor processing time and resource usage
* **Error handling**: View any processing errors or warnings

***

## Step 4: Quality Assessment and Review

After transformation, the system automatically performs **quality validation** on your extracted knowledge graph.

### Understanding Quality Metrics

The quality system provides:

* **Overall Score**: 0-100 scale with letter grades (A-F)
* **Violation Summary**: Categorized by severity (Error, Warning, Info)
* **Rule Types**: Format rules, business rules, and completeness checks
* **Detailed Reports**: Entity-level and property-level validation results

### Quality Rules Include:

* **Format Rules**: Data type validation, format compliance
* **Business Rules**: Domain-specific constraints and validations
* **Completeness Rules**: Required property validation, relationship verification
* **Consistency Rules**: Cross-entity validation and constraint checking

### Quality Decision Points

Based on the quality results, you have several options:

**High Quality (Score ≥ 90, No Errors)**:

* ✅ **Approve and Proceed**: Automatically move to merge process
* Quality is sufficient for production use

**Moderate Quality (Score 70-89)**:

* ⚠️ **Manual Review Required**: Examine violations and make informed decision
* Consider acceptable risk vs. data value
* Add feedback comments for future improvements

**Low Quality (Score \< 70)**:

* ❌ **Reject and Retry**: Fix source data or adjust ontology
* Address critical errors before proceeding
* Provide detailed feedback for system learning

***

## Step 5: Provide Feedback and Iterate

The quality assessment stage allows you to provide valuable feedback to improve future processing.

### Feedback Options

1. **Approval with Comments**: Explain why you're accepting despite issues
2. **Rejection with Reasons**: Detail what needs to be fixed
3. **Quality Suggestions**: Recommend improvements for similar data

### Retry Process

If you reject the quality results:

1. **Address root causes**: Fix source documents or adjust ontology
2. **Re-run transformation**: Process the corrected data
3. **Review improved results**: Check if quality issues are resolved

***

## Step 6: Merge Process and Conflict Resolution

If quality is approved, the system proceeds to the **Merge** page to integrate your new data with existing production data.

### Automatic Conflict Detection

The system automatically detects potential conflicts:

* **Duplicate entities**: Same entity with different properties
* **Conflicting values**: Different values for the same property
* **Relationship conflicts**: Inconsistent relationship definitions
* **Schema mismatches**: Entity structure differences

### Human-in-the-Loop Resolution

When conflicts are detected, **human involvement is required**:

#### Conflict Resolution Options

1. **Accept New Data**: Use the newly extracted information
2. **Keep Existing**: Maintain current production data
3. **Merge Values**: Combine information from both sources
4. **Custom Resolution**: Create a hybrid solution

#### Learning Comments

For each resolution, you can provide **learning comments**:

* Explain why you chose a particular resolution
* Provide context for similar future conflicts
* Help the system learn your preferences
* Enable automated resolution of similar issues

**Example Learning Comment**:

```
"When company names differ only in legal entity type (Inc. vs. Corp.), 
prefer the more recent document as it's likely more current."
```

### Future Automation

Your learning comments enable the system to:

* Automatically resolve similar conflicts in the future
* Apply your business rules consistently
* Reduce manual intervention over time
* Improve overall system intelligence

***

## Step 7: Final Persistence and Versioning

After all conflicts are resolved, the final step persists your data to the production database.

### Data Persistence

* **Clean Integration**: Resolved data is stored in production Neo4j
* **Relationship Creation**: All entity relationships are properly linked
* **Index Updates**: Search indexes are updated for optimal performance

### Versioning Strategy

Graphora implements a sophisticated versioning system:

#### Version Management

* **New Nodes**: Completely new entities get version 1
* **Updated Nodes**: Modified entities get incremented versions
* **Historical Preservation**: Previous versions are retained, not deleted
* **Version Marking**: Old nodes are marked as previous versions

#### Benefits of Versioning

1. **Audit Trail**: Complete history of all changes
2. **Rollback Capability**: Ability to revert to previous versions
3. **Temporal Queries**: Query data as it existed at specific points in time
4. **Change Tracking**: Monitor how your knowledge graph evolves

#### Version Resolution Example

```
Original Node: Company "ABC Corp" (v1)
├── Updated to: Company "ABC Corporation" (v2) ← Current
└── Previous: Company "ABC Corp" (v1) ← Retained but marked as previous
```

***

## Step 8: Monitoring and Maintenance

After successful data integration, monitor your knowledge graph's ongoing performance.

### Available Monitoring

* **Quality Trends**: Track quality improvements over time
* **Conflict Patterns**: Identify common conflict types
* **Processing Metrics**: Monitor transformation performance
* **Data Growth**: Track knowledge graph expansion

### Best Practices

1. **Regular Quality Reviews**: Periodically review quality thresholds
2. **Ontology Evolution**: Update schemas as your domain evolves
3. **Feedback Analysis**: Review learning comments for optimization opportunities
4. **Performance Monitoring**: Watch for processing bottlenecks

***

## Troubleshooting Common Issues

### Configuration Problems

* **Database Connection Failed**: Verify URI, credentials, and network access
* **API Key Invalid**: Check Gemini API key format and permissions
* **Permission Errors**: Ensure database user has required permissions

### Schema Issues

* **Validation Errors**: Review YAML syntax and property naming (use camelCase)
* **Missing Relationships**: Ensure all entity relationships are properly defined
* **Quality Rule Conflicts**: Check for contradictory validation rules

### Processing Problems

* **Transformation Stuck**: Check document format and file size limits
* **Low Quality Scores**: Review source data quality and ontology alignment
* **Memory Issues**: Consider breaking large documents into smaller chunks

### Merge Conflicts

* **Too Many Conflicts**: Review entity matching criteria and uniqueness rules
* **Resolution Failures**: Ensure learning comments are clear and actionable
* **Performance Issues**: Monitor conflict resolution processing time

***

## Getting Help

If you encounter issues during any step:

1. **Check Documentation**: Review relevant sections of this guide
2. **Quality Reports**: Use detailed quality violation reports for debugging
3. **System Logs**: Monitor processing logs for error details
4. **Support Contact**: Reach out to [support@graphora.io](mailto:support@graphora.io) for assistance

***

## Next Steps

Once you've completed your first successful workflow:

* **Explore Advanced Features**: Learn about bulk processing and automation
* **API Integration**: Consider using the [GraphoraClient library](/client/introduction) for programmatic access
* **Quality Optimization**: Fine-tune your ontology and quality rules
* **Scale Your Operations**: Process larger document sets and complex ontologies

<Card title="Python Client Library" icon="code" href="/client/introduction">
  Learn how to integrate Graphora programmatically using our Python client library.
</Card>

***

*This guide covers the complete end-to-end workflow for Graphora platform users. Each step builds on the previous one to ensure a smooth and successful knowledge graph creation experience.*