LLM/ChatGPT AI for Sentiment and Topic Analysis
Client
The Forbes Funds is a Pittsburgh, PA based nonprofit that advances the well-being of its region by helping human service and community-based nonprofits build their management capacity and increase the impact of their mission work.
Problem
The Forbes Funds offered several open-ended responses as part of their survey. As a community organization, surveys are critical to the work of the Forbes Funds. It’s imperative that they are able to summarize and explain resident sentiments to key stakeholders and policymakers. ChatGPT and OpenAI make it cheap and easy to quickly analyze sentiments, and then feed them into a dashboard for reporting.
Solution
We coded a custom integration with ChatGPT that programmatically read in survey responses.
While much of the Forbes Funds data work is public because they are a nonprofit, certain information from their surveys must be kept private because it contains identifying information. Therefore, for the sake of anonymization, we removed most of the fields in the input dataset, and also removed any connections to live data sources. Typically, the script would ingest the entire survey dataset from Google Drive, where it is auto-updating in real time from a form. Then it would perform this sentiment analysis as well as other cleaning operations, and then send the data to their live database that is connected to Tableau. The script would run every hour to capture new responses and keep the dashboard up to date. For this case study, we stripped out the extra stuff and only displayed the relevant code for the sentiment analysis piece. We also took a sample of the survey responses.
