At Zencity, we're committed to providing the most accurate and relevant data insights for local governments. We're excited to announce an update to our sentiment analysis model that focuses specifically on measuring sentiment in relation to local government- starting April 24, 2025. This new model enhances our ability to capture and analyze community opinions on matters directly related to municipal governance and services.
New model vs. old model
Until today, our sentiment model was discerning how satisfied the author of a post or comment was, with the issue at hand. The limitations of this approach was that it was indifferent to the context of the issue or to the relation between that issue and the relevant local government.
Our updated sentiment model now specifically asks: "Would the author of this post or comment express support for or opposition to your organization's measure or action on this issue?" This targeted approach allows us to more accurately gauge public sentiment on matters directly relevant to your organization.
Key Features of the New Model
- Contextual Understanding: Our algorithm has been trained to recognize and interpret the context of a post, comments and engagements. For example, it can understand when a supportive comment may actually be negative in regards to local government, since the support it is conveying is directed to a city-criticizing comment.
- Improved Accuracy: By narrowing the focus to local government-related sentiments, we've enhanced the model's precision in classifying relevant content.
Your Role in Improving the Model
Your feedback remains crucial. If you encounter any misclassified items in your dashboard, please use the feedback feature to suggest the appropriate sentiment label. This input helps us continuously refine and improve our local government-focused sentiment analysis. In fact, we’ve used our client’s feedback to train this new model!
FAQ
| Question | Suggested Answer |
| Why did you change the model? | Clients told us they needed sentiment that reflects public alignment or pushback on their actions—not just how people feel generally. This new model delivers that. |
| What’s different from before? | Old sentiment = general tone in a certain text. New sentiment = support or opposition to the local gov’s actions on that topic. |
| Will this change my historic sentiment data? | No, the update applies going forward. You’ll start to see changes reflected in new content classifications. |
| Why is this more accurate? | Because it narrows the scope. We’re no longer trying to judge any emotion—we’re focused on what matters to public leaders: alignment or friction with your work. |
| Can I give feedback? | Yes, and we encourage it. The feedback feature in your dashboard helps us fine-tune the model based on real-world edge cases. |
| What’s an example of this in action? | Imagine someone says “Finally, someone’s standing up to the city!” That used to look positive. Now, it’s flagged as opposition to the city’s action—because that’s what it is. |
Comments
0 comments
Please sign in to leave a comment.