Introduction
The GPT Comment Analysis Toolkit within the Public Input platform is a sophisticated, question-specific tool that revolutionizes the way comments are analyzed and processed. It offers an efficient, accurate, and user-friendly approach to understanding public feedback for each individual question in your project.
Key Features
Theme identification: The GPT comment analysis tool evaluates comments in response to a specific prompt or question to identify common themes.
Customizable Tags: Flexibility to modify AI-suggested themes to fit the specific nuances of your question or project.
Efficient Comment Tagging: AI accelerates the tagging of comments relevant to specific questions.
Integration with existing visualization and reporting tools: Utilize bubble clouds and other visualization tools to discern patterns and insights for each question.
Using the tool
Toolkit Location and Scope
Found in Project Admin Interface: This toolkit is located within the 'Comment Analysis' tab of the project admin interface.
Focused on Individual Questions: It is designed to analyze comments on a per-question basis, not for all comments across various questions.
Accessing the tool and identifying themes
Accessing the Tool: Navigate to the 'Comment Analysis' tab within the project admin interface of your specific project.
Select an individual question in the left side bar: Upon selecting a question, the right pane will show that question's comments and a top bar that includes an "AI Comment Analysis" tab.
Identify themes with AI: After selecting the AI Comment analysis tab, click the "Run AI Comment Analysis" button to run the AI analysis on comments.
Reviewing themes and editing
Human Oversight and Control: Review the AI-suggested themes. Edit, delete, or add themes to tailor the analysis to the specific context of the question.
Applying themes as tags to comments
AI-Assisted Tagging of Comments: Apply the finalized themes as tags to each comment corresponding to the question.
Review and Adjustment of Tags: Ensure accuracy by reviewing and adjusting tags on each comment.
Using insights from AI-generated tags
Tag Cloud Visualization
Once comments are tagged, use the bubble cloud feature to visually explore the themes and patterns emerging from the comments.
Bubble clouds can be added to your dynamic reports. Simply mouse over the visualization, and a menu will appear at the top right corner. Select the report from the dropdown list, and the chart will be added to the report, defaulting to the top. Open your report to drag the visualization to a specific section.
Filtering and Exploration
Easily filter and explore comments based on their tags to understand the public sentiment on specific aspects of the question. Clicking on any tag cloud element will filter your comments to that specific tag:
To remove the filter, click the "x" icon next to the filter below the tag cloud visualization.
Batch Response Capabilities
Respond to groups of comments with similar themes in a batch, enhancing efficiency and consistency in public engagement.
Select multiple comments by checking the comments you'd like to reply to, or by clicking the select all checkbox at the top of the comment feed.
From here you can click "Reply to multiple" to respond to each comment. If we have an email for a commenter, your response will be emailed to the participant if you select the "Send user an email" option:
Your sent reply will appear with your comment records in the admin UI, comment exports, and if you selected "Send user an email notification", users who provided their email will be notified of the new reply. Learn more about replying to comments here.
Pro tips:
Tags should signal a specific perspective. When creating tags, consider including a “position” or sentiment, for example:
“Supports bike lanes” and “Opposes bike lanes” are better than “Bike lanes”
You can multi-task. While the system is analyzing comments for themes, you can start the process for another question.
Add your own tags. You may have additional context that the comments can’t provide to GPT. After the initial theme identification step, add your own tags before running the “Apply tags” step.
See also: