MessageNodeEditor Documentation

Overview

The MessageNodeEditor is a sophisticated component within the Vantage analytics and data platform that is designed for configuring message processing workflows. It facilitates the parsing and analysis of messages, enabling businesses to extract valuable insights such as sentiment, intent, and topics present in the messages received. The MessageNodeEditor integrates seamlessly with various upstream data sources and provides a user-friendly interface for customizing how messages are analyzed and categorized.

Purpose

The primary purpose of the MessageNodeEditor is to:

• Define how messages are parsed and structured from input data sources.
• Utilize AI and NLP capabilities to analyze messages for sentiment, intent, and other relevant metrics.
• Customize the categorization of messages based on configurable settings.
• Output the results into a designated format for further processing or storage.

Settings

Below are detailed explanations of each setting available within the MessageNodeEditor:

1. Message Column

• Name: messageColumn
• Input Type: Dropdown
• Description: This setting allows users to specify the column in the upstream data source that contains the message text. The analyzer will operate on messages from this column.
• Default Value: body

2. Sender Column

• Name: senderColumn
• Input Type: Dropdown
• Description: Users can select the column that represents the sender of the message. This is useful for categorizing and understanding the source of messages.
• Default Value: from

3. Subject Column

• Name: subjectColumn
• Input Type: Dropdown
• Description: This allows users to map the column that contains the subject of the message, aiding in further categorization and context analysis.
• Default Value: subject

4. Channel Column

• Name: channelColumn
• Input Type: Dropdown
• Description: Users can select the column used to indicate the communication channel (e.g., email, SMS, etc.). It can help in analyzing message trends based on channels.
• Default Value: (Auto-detect)

5. Categories

• Name: categories
• Input Type: Array of Objects
• Description: Users can define custom categories for messages. Each category includes a label and associated keywords, allowing for tailored categorization.
• Default Value: A single default category labeled as "General".

6. Default Category

• Name: defaultCategory
• Input Type: String
• Description: Specifies the default category assigned to messages if no custom categories match. This ensures that all messages are categorized, even when they do not meet any specific criteria.
• Default Value: "General"

7. Analysis Type

• Name: analysisType
• Input Type: Dropdown
• Description: This determines the type of AI analysis applied to the messages. Options include sentiment, intent, and custom analyses, providing flexibility based on user needs.
• Default Value: sentiment

8. Custom Prompt

• Name: customPrompt
• Input Type: String
• Description: Users can define a custom prompt for the AI analysis. This allows for tailored responses in scenarios where generic analysis does not suffice.
• Default Value: "" (Empty)

9. Output Column

• Name: outputColumn
• Input Type: String
• Description: Specifies the name of the column where the output of the AI analysis will be stored in the dataset.
• Default Value: ai_analysis

10. Batch Size

• Name: batchSize
• Input Type: Numeric (range slider)
• Description: This controls the number of messages processed in each batch during AI analysis, impacting performance and memory usage.
• Default Value: 25

11. Tone

• Name: tone
• Input Type: Button group
• Description: Defines the tone in which the responses will be structured - options include professional, friendly, empathetic, and formal.
• Default Value: professional

12. Response Length

• Name: responseLength
• Input Type: Dropdown
• Description: Specifies the length of the AI-generated responses to be brief, standard, or detailed.
• Default Value: standard

How It Works

The MessageNodeEditor component manages multiple aspects of message processing through a series of sub-editors that allow for detailed configuration:

1. Column Mapping: Users can map relevant columns from upstream data, ensuring that each message and its context are accurately processed.
2. AI Analysis: Users can select different types of analysis based on the messages being analyzed. Through AI integration, the system analyzes the messages for desired attributes such as sentiment and intent.
3. Category Management: Users can define a set of categories that apply to messages. The system will check incoming messages against these categories.
4. Customization via Prompts: If the default analysis is insufficient, users can customize prompts for specific scenarios, resulting in tailored AI responses and analysis outputs.

Data Expectations

The MessageNodeEditor expects input data to be structured in a tabular format, with relevant columns conforming to the fields defined in the settings (message, sender, subject, channel). It processes the data according to the user-defined configuration, performing tasks like parsing, categorization, and analysis, before outputting the results into the designated output column.

Use Cases & Examples

Use Cases

1. Customer Support Automation: A company can set up the MessageNodeEditor to analyze incoming customer queries for sentiment and category. This enables automated routing of support requests based on urgency and customer sentiment.

2. Marketing Insights: Businesses can utilize the component to categorize and analyze feedback from marketing campaigns. By understanding the sentiment associated with different campaigns, they can pivot strategies in real-time.

3. Social Media Analysis: The tool can parse messages from various social media platforms to gauge public opinion, detect trending topics, and derive insights into customer engagement.

Example Configuration

Use Case Scenario: A customer support team wants to categorize incoming messages based on urgency and topic.

Sample Configuration Data:

{
  "messageColumn": "body",
  "senderColumn": "from",
  "subjectColumn": "subject",
  "channelColumn": "channel",
  "categories": [
    { "label": "Urgent", "keywords": "help, urgent, ASAP" },
    { "label": "General", "keywords": "info, question" }
  ],
  "defaultCategory": "General",
  "analysisType": "topics",
  "outputColumn": "ai_analysis,
  "batchSize": 25,
  "tone": "professional",
  "responseLength": "brief"
}

In this configuration:

• The team is equipped to parse the body of incoming messages, tagging them based on urgency and topic.
• They can quickly respond to pressing customer needs while categorizing all inbound queries for future analysis.
• The output will be stored in ai_analysis, allowing for efficient reporting and tracking.