Comment Topic and Sentiment Upgrade

We’ve upgraded the sentiment and topic classification models behind Text Analytics to deliver more accurate and insightful results from your survey comments. This update improves how comments are tagged and summarized, helping you get even more value from open-text feedback.

What’s New:

• Powered by Large Language Models (LLMs): The new model uses LLM technology to improve sentiment and topic tagging accuracy.
• Improved context-awareness: Survey questions are now included as context when classifying comments, leading to more accurate and relevant tags.
• Redefined 'Neutral' sentiment: - Previously: Comments that didn’t clearly sound positive or negative were labeled as Neutral. Now: Neutral includes both truly neutral and mixed sentiment comments, those with a mix of positive and negative language.
• Improved accuracy overall: Better handling of short or ambiguous comments (e.g., “flexible working”). More accurate AI summaries due to improved underlying tagging.

Check out the Text Analytics in Comments Reporting article to learn more.