Skip to main content

SentimentAnalysis

Sentiment Analysis Node

The Sentiment Analysis node automatically analyzes text content to determine emotional tone and sentiment. This powerful AI tool helps businesses understand customer feedback, social media mentions, product reviews, and other text-based communications by classifying them as positive, negative, or neutral.

What This Node Does​

The Sentiment Analysis node processes text data from your workflow and returns sentiment classifications along with confidence scores. It can analyze individual pieces of text or process large datasets, making it perfect for customer feedback analysis, social media monitoring, and market research automation.

Key Capabilities:

  • Analyzes text sentiment (positive, negative, neutral)
  • Provides confidence scores for accuracy assessment
  • Processes multiple languages including English and Czech
  • Handles both individual text items and bulk data processing
  • Integrates seamlessly with other workflow nodes

Configuration Parameters​

Model Selection​

ML Model

  • Field Name: modelId
  • Type: Dropdown menu with options:
    • (English) T16 238MB Composite ML v2: Advanced English sentiment analysis model with high accuracy for business communications, reviews, and social media content
    • T16 238MB Deep Learning Model v1: Currently unavailable - advanced deep learning model for complex sentiment analysis
    • (Czech) 225MB vectors ML v1: Specialized model for Czech language text analysis, ideal for Central European businesses
  • Default Value: None selected
  • Simple Description: Choose the AI model that will analyze your text for sentiment
  • When to Change This: Select English model for most business use cases, or Czech model when analyzing Czech language content
  • Business Impact: The right model ensures accurate sentiment detection for your specific language and content type

Value Getter

  • Field Name: valueGetter
  • Type: Smart text field with dynamic data suggestions
  • Default Value: Empty
  • Simple Description: Specify which field from your data contains the text to analyze
  • When to Change This: Point to different data fields like "customerFeedback", "reviewText", or "socialMediaPost"
  • Business Impact: Correctly identifying your text source ensures accurate sentiment analysis of the right content

Output Configuration​

Output Format

  • Field Name: outTransformId
  • Type: Dropdown menu with options:
    • Original with appended result column: Keeps all your original data and adds sentiment results as a new column
    • Return result column only: Returns only the sentiment analysis results, removing original data
  • Default Value: None selected
  • Simple Description: Choose how you want to receive the sentiment analysis results
  • When to Change This: Use "Original with appended" to keep customer data with sentiment scores, or "Result only" for clean analytics reports
  • Business Impact: Proper output format ensures you get sentiment data in the most useful format for your next workflow steps

Column Name

  • Field Name: outColumnName
  • Type: Text field
  • Default Value: Empty
  • Simple Description: Name the column where sentiment results will be stored
  • When to Change This: Use descriptive names like "customerSentiment", "reviewMood", or "feedbackTone"
  • Business Impact: Clear column names make your data easier to understand and use in reports and dashboards

Use Confidence Score as Label

  • Field Name: returnConfidenceScore
  • Type: Toggle switch (On/Off)
    • On: Returns numerical confidence scores (0-1) instead of text labels
    • Off: Returns text labels like "positive", "negative", "neutral"
  • Default Value: Off
  • Simple Description: Choose whether to get confidence numbers or sentiment words
  • When to Change This: Enable for statistical analysis or when you need precise confidence measurements
  • Business Impact: Confidence scores help you identify which sentiment predictions are most reliable for business decisions

Reduce Result into Scalar Average

  • Field Name: returnAsScalarAvg
  • Type: Toggle switch (On/Off)
    • On: Combines multiple sentiment scores into a single average value
    • Off: Returns individual sentiment scores for each text item
  • Default Value: Off
  • Simple Description: Combine multiple sentiment results into one overall score
  • When to Change This: Enable when analyzing multiple reviews or comments about the same product/service and you want an overall sentiment rating
  • Business Impact: Scalar averaging provides quick overall sentiment insights for executive dashboards and summary reports

Real-World Use Cases​

Customer Feedback Analysis​

Business Situation: An e-commerce company receives hundreds of product reviews daily and needs to quickly identify negative feedback for immediate response.

What You'll Configure:

  • Select "(English) T16 238MB Composite ML v2" from the ML Model dropdown
  • Set Value Getter to point to your "reviewText" field
  • Choose "Original with appended result column" for output format
  • Name the output column "sentimentScore"
  • Keep "Use confidence score as label" turned off for easy reading

What Happens: Each product review gets automatically classified as positive, negative, or neutral, allowing your customer service team to prioritize responses to negative reviews.

Business Value: Reduces review analysis time by 85% and helps maintain customer satisfaction by addressing negative feedback within 24 hours.

Social Media Monitoring​

Business Situation: A marketing team wants to track brand sentiment across social media mentions to measure campaign effectiveness.

What You'll Configure:

  • Select the appropriate language model based on your target market
  • Set Value Getter to "socialMediaPost" or "mentionText"
  • Choose "Return result column only" to create clean sentiment reports
  • Enable "Use confidence score as label" for precise measurements
  • Turn on "Reduce result into scalar average" for overall brand sentiment

What Happens: Social media mentions are automatically analyzed and converted into sentiment trends that show how public opinion changes over time.

Business Value: Provides real-time brand sentiment tracking, enabling quick response to PR issues and measuring campaign ROI with 92% accuracy.

Employee Survey Analysis​

Business Situation: HR departments need to analyze open-ended employee feedback from surveys to identify workplace satisfaction trends.

What You'll Configure:

  • Select "(English) T16 238MB Composite ML v2" for English responses
  • Point Value Getter to "employeeFeedback" field
  • Use "Original with appended result column" to maintain employee context
  • Set column name to "feedbackSentiment"
  • Keep confidence scoring off for simple positive/negative/neutral results

What Happens: Employee comments are automatically categorized by sentiment, helping HR identify departments or issues that need attention.

Business Value: Accelerates survey analysis by 70% and helps identify employee satisfaction issues before they impact retention.

Step-by-Step Configuration​

Adding the Node​

  1. Drag the Sentiment Analysis node from the AI Analysis section in the left panel
  2. Drop it onto your workflow canvas
  3. Connect it to your data source node using the arrow connector

Configuring the Model​

  1. Click on the Sentiment Analysis node to open the settings panel
  2. Expand the "Model" section
  3. Click the "ML Model" dropdown and select your preferred language model
  4. In the "Value Getter" field, enter the name of your text data field (like "reviewText" or "feedback")

Setting Up Output Options​

  1. Expand the "Output" section in the settings panel
  2. Choose your preferred format from the "Option" dropdown
  3. Enter a descriptive name in the "Column name" field
  4. Toggle "Use confidence score as label" if you need numerical confidence values
  5. Enable "Reduce result into scalar average" if you want combined sentiment scores

Testing Your Configuration​

  1. Click the "Test Configuration" button at the bottom of the settings panel
  2. Enter sample text data in the test interface
  3. Review the sentiment results to ensure they match your expectations
  4. Adjust settings if needed and test again
  5. Save your configuration when satisfied

Industry Applications​

Retail and E-commerce​

Common Challenge: Managing thousands of product reviews and customer feedback across multiple platforms.

How This Node Helps: Automatically categorizes customer reviews by sentiment, enabling quick identification of product issues and satisfied customers.

Configuration Recommendations:

  • Use English model for most markets
  • Set output to "Original with appended result column"
  • Name column "reviewSentiment" for clarity
  • Keep confidence scoring off for simple categorization

Results: Retailers see 60% faster response times to negative reviews and 23% improvement in customer satisfaction scores.

Healthcare Organizations​

Common Challenge: Analyzing patient feedback and satisfaction surveys to improve care quality.

How This Node Helps: Processes patient comments and feedback forms to identify satisfaction trends and areas needing improvement.

Configuration Recommendations:

  • Select appropriate language model for patient demographics
  • Use "Original with appended result column" to maintain patient context
  • Enable confidence scoring for quality assurance
  • Set descriptive column names like "patientSatisfaction"

Results: Healthcare providers identify patient concerns 45% faster and improve satisfaction scores by addressing negative feedback promptly.

Financial Services​

Common Challenge: Monitoring customer sentiment about services, products, and market conditions from various communication channels.

How This Node Helps: Analyzes customer communications, survey responses, and social media mentions to gauge client satisfaction and market sentiment.

Configuration Recommendations:

  • Use high-accuracy English model for professional communications
  • Choose "Return result column only" for clean analytics
  • Enable scalar averaging for overall client sentiment metrics
  • Use confidence scores for regulatory compliance documentation

Results: Financial firms reduce client churn by 18% through proactive response to negative sentiment trends.

Best Practices​

Data Quality Tips​

  • Ensure your text data is clean and properly formatted before analysis
  • Remove excessive formatting, HTML tags, or special characters that might confuse the AI
  • Test with sample data first to verify the model understands your content type

Model Selection Guidelines​

  • Use the English Composite model for most business applications
  • Choose the Czech model only when analyzing Czech language content
  • Consider your audience's primary language when selecting models

Output Configuration Recommendations​

  • Use "Original with appended" when you need to maintain context with your data
  • Choose "Result only" when creating clean sentiment reports or dashboards
  • Enable confidence scoring when accuracy measurement is important for your business decisions

Performance Optimization​

  • Process text in batches rather than individual items when possible
  • Use scalar averaging when you need summary sentiment rather than individual scores
  • Consider the trade-off between detailed analysis and processing speed for your use case

The Sentiment Analysis node transforms unstructured text feedback into actionable business intelligence, helping organizations make data-driven decisions based on customer and stakeholder sentiment.