Three years ago we asked ourselves, “How can natural language processing (NLP) and machine learning technologies help digital democracy platforms?” CitizenLab developed a feature that realizes just that.
In 2022, we released our “Insights” feature to help our clients analyze and categorize community feedback on their community engagement platforms. This is particularly valuable for local governments that gather text-heavy engagements on a wide range of topics. With the help of auto-tagging, community engagement managers can easily process thousands of qualitative inputs within minutes, freeing up valuable time and resources for deeper and more continuous engagement.
We believe this is just the beginning. Community engagement is more than just processing community input into categories or counting the number of posts. That’s why we have also built a visual keyword map using NLP and artificial intelligence (AI) capabilities built into our platform. The keyword map lets you go beyond auto-tagging and enables you to visually uncover community insights.
Here are four ways cities can leverage our Insights tool
AI helps you process qualitative inputs
With the help of AI and NLP, our tool reads the deeper meaning and context behind each post. It can determine whether a particular post is semantically related to a category and provide recommendations.
By clicking on the button – community managers can get recommendations of the posts’ relevant categories
This helps our community engagement managers efficiently process text input that they gathered from their community. For engagement projects with hundreds of inputs, it can be resource-intensive to process, analyze, and uncover the insights. With our platform, community engagement managers get recommendations of the posts’ relevant categories with a click of a button.
Visually explore what people are talking about
We use an NLP model to analyze the most popular keywords or concepts that are being discussed within the community. The most popular keywords are then visualized on a map relative to their importance and clustered together with keywords that appear together repeatedly.
Community managers can have an overview of all the trending keywords within the community
The visual representation helps community managers quickly understand what people are discussing within the community and group the themes in order of importance. The community manager can then quickly see what the trends, changes, and gaps are in the current project and decide whether a detailed analysis is necessary for the next steps.
Remove bias from data sorting
The map shows the keywords that are discussed frequently together in the same area of the map. This group of keywords shows the key topics or themes discussed within the community. The community managers can then interact with the map to select keywords, create tags, filter, or search posts.
Community managers can interact with keywords, create tags, filter and search for posts easily
This helps community managers process input intuitively on the map instead of going through individual posts on a tedious spreadsheet. More importantly, there is no longer the need for community managers to have a predefined list of categories which removes any potential for bias and focuses on what the community is truly identifying as priorities.
AI offers an inclusive overview of the community
The more frequently a keyword is mentioned, the more central and visually prominent it becomes, which helps community managers easily spot key topics of interest. Beyond showing what the main trending topics are, our keyword map also clusters keywords that may not be part of the larger discussion. For instance, one client’s urban planning project showed that the majority interest was around parks and nature, but a related group of keywords about student life was also clustered at the corner of the keyword map. In turn, this could influence the urban planning project, as well as how and for whom it was designed. By showing both highly-trending as well as ancillary topics, our platform helps community engagement managers be more inclusive in their analysis.
Community managers can easily spot minority interests within the community on the map
This helps community managers easily spot minority interests within the community and further analyze what their needs are.
What’s next for CitizenLab’s Community Insights?
Our Insights feature currently works in 16 languages, including; English, Dutch, German, French, Spanish, Arabic, and Polish. We are working on improving the accuracy of the current languages and adding support for more languages. Additionally, we are exploring how we can leverage the tool further by extending it to surveys, comment analysis, and other qualitative inputs the platform collects.
Look out for more exciting updates in the coming weeks/months!
