Characterizations

The Characterizations capability allows administrators to define custom categorization systems for incoming questions, enabling systematic classification across multiple dimensions for better analysis and reporting.

Overview

Characterizations provide a flexible framework for categorizing and analyzing user interactions across multiple dimensions. Each characterization represents a specific aspect of questions or interactions that your organization wants to track and analyze. This system automatically applies these categorizations to every incoming question, providing rich metadata for reporting and insights.

Key Capabilities

Characterization Management

Create Characterization

The "Create Characterization" button opens a wizard to define a new categorization dimension:

Required Information:

• Characterization Name: Descriptive name for the categorization dimension (e.g., "Product Type", "User Sentiment", "Language")
• Description: Brief explanation of what this characterization tracks
• Options: List of possible values for this characterization
• Other Category: Always includes an "Other" option as a catch-all for unmatched items

Characterization Structure: Each characterization consists of:

• Characterization Name: The main categorization dimension (e.g., "Product Type")
• Options: Specific values within that category (e.g., "Hardware", "Software")
• Descriptions: Detailed explanations of each option
• Other: Default catch-all option for unmatched items

Characterization Table

The main table displays existing characterizations:

Column	Description
Name	Characterization dimension name
Options Count	Number of defined options for this characterization
Status	Active/Inactive status
Last Modified	Date of last update
Actions	Edit, duplicate, or delete options

Common Characterization Examples

Product-Related Characterizations

Product Type

Categorize questions based on the type of product being discussed:

Options:

• Hardware: Physical products, devices, equipment
• Software: Applications, programs, digital tools
• Services: Consulting, support, implementation
• Firmware: Embedded software in hardware devices
• Other: Any product type not covered above

Product Component

Classify questions by specific product components or features:

Options:

• User Interface: Frontend, UI/UX, user experience
• Backend: Server-side, database, APIs
• Integration: Third-party connections, APIs, connectors
• Performance: Speed, optimization, scalability
• Security: Authentication, authorization, data protection
• Other: Components not covered above

User Experience Characterizations

User Sentiment

Track the emotional tone or satisfaction level of user inquiries:

Options:

• Positive: Satisfied, happy, complimentary
• Neutral: Factual, informational, routine
• Frustrated: Confused, annoyed, experiencing issues
• Urgent: Time-sensitive, critical, emergency
• Other: Sentiment not covered above

User Type

Categorize users based on their role or relationship:

Options:

• Customer: Paying customers or clients
• Prospect: Potential customers evaluating products
• Partner: Business partners or resellers
• Internal: Company employees or team members
• Other: User types not covered above

Technical Characterizations

Language

Track the language of incoming questions:

Options:

• English: Primary language for most interactions
• Spanish: Spanish language support
• French: French language support
• German: German language support
• Other: Additional languages not covered above

Technical Complexity

Assess the technical depth of questions:

Options:

• Basic: Simple questions, general information
• Intermediate: Moderate technical depth, specific features
• Advanced: Complex technical issues, deep troubleshooting
• Expert: Highly technical, architectural questions
• Other: Complexity levels not covered above

Configuration Process

Step 1: Define Characterization

1. Access Characterization Page: Navigate to Configuration > Characterizations
2. Create New: Click "Create Characterization" button
3. Basic Information: Enter name and description
4. Define Options: List all possible values with descriptions

Step 2: Configure Options

For each option within a characterization:

Required Fields:

• Option Name: Short, clear label for the option
• Description: Detailed explanation of what this option covers
• Keywords: Optional terms that might trigger this classification
• Priority: Order of importance for display and analysis

Option Management:

• Add Options: Dynamically add new options as needed
• Edit Existing: Modify option names, descriptions, or keywords
• Reorder: Change the priority or display order of options
• Archive: Temporarily disable options without deleting

Step 3: Deliver Configuration

If you would like to implement characterizations please email us at info@ept.ai with the channel id you'd like the characterizations applied to and the json characterization using the format below.

Export Format:

{
  "Product Type": {
    "Hardware": "Physical products, devices, equipment",
    "Software": "Applications, programs, digital tools",
    "Services": "Consulting, support, implementation",
    "Firmware": "Embedded software in hardware devices",
    "Other": "Any product type not covered above"
  },
  "User Sentiment": {
    "Positive": "Satisfied, happy, complimentary",
    "Neutral": "Factual, informational, routine",
    "Frustrated": "Confused, annoyed, experiencing issues",
    "Urgent": "Time-sensitive, critical, emergency",
    "Other": "Sentiment not covered above"
  },
  "Language": {
    "English": "Primary language for most interactions",
    "Spanish": "Spanish language support",
    "French": "French language support",
    "German": "German language support",
    "Other": "Additional languages not covered above"
  }
}

Structure Explanation:

• Top Level: Each key is a characterization name (e.g., "Product Type", "User Sentiment")
• Second Level: Each characterization contains a list of options
• Option Level: Each option has the option name as the key and description as the value
• Always Include "Other": Each characterization must have an "Other" option as a catch-all

Automatic Classification

Real-Time Processing

Every incoming question is automatically analyzed and classified:

Classification Process:

1. Question Analysis: AI analyzes the content and context
2. Pattern Matching: Compares against defined keywords and descriptions
3. Classification: Assigns appropriate options from each characterization
4. Confidence Scoring: Provides confidence level for each classification
5. Fallback Handling: Uses "Other" category when no clear match exists

Reporting and Analytics

CSV Export Integration

Characterization data is included in all CSV downloads from the reporting section.

Best Practices

Characterization Design

Naming Conventions

1. Clear and Descriptive: Use names that clearly indicate what is being categorized
2. Consistent Format: Maintain consistent naming patterns across characterizations
3. Business Language: Use terminology familiar to your team and stakeholders
4. Avoid Ambiguity: Ensure names don't overlap or create confusion

Option Structure

1. Mutually Exclusive: Options should not overlap or conflict
2. Comprehensive Coverage: Include all major categories within each dimension
3. Logical Grouping: Organize options in logical, intuitive order
4. Other Category: Always include "Other" as a catch-all option

Description Quality

1. Clear Definitions: Provide unambiguous descriptions for each option
2. Examples: Include specific examples where helpful
3. Keywords: Add relevant keywords for better AI classification
4. Regular Updates: Keep descriptions current with product changes

Related Features

• Reporting - Access characterization data in CSV exports
• Responses - View how characterizations affect response quality
• Knowledge Sources - Characterizations help organize knowledge by category
• Channels - Characterizations apply across all configured channels