Inspiration
I've lived in San Francisco since 2013, and it's my favorite city. Recently I've noticed there has been an increase in shootings and even murders right outside my apartment, and robbery at gunpoint at my local convenience store. A potential co-founder as well as a close friend both decided to not live in SF because their wives feel unsafe walking around the city alone. I'm hoping that since "what gets measured, gets improved", that if we could measure problems (like gun violence), without bias, and measure discussions of decision-makers, that we would prioritize safety in SF.
What it does
It measures 911 calls that are related to shootings and people carrying guns, and then goes through every decision-making meeting transcript, and checks whether that issue is being discussed by our local politicians (in Board of Supervisors meetings).
Please take a look at the raw data here - https://docs.google.com/spreadsheets/d/1zwmLckPRycprgdSXXYHObW3BlpbNpxtbYwwJI6aJA4Y/edit#gid=133017521 If you notice anything off, please let me know and I'll update it.
How we built it
Simple data pipeline in Python.
Challenges we ran into
The pipeline is very slow, not sure how to make this faster. Claude and/or knowledge of Claude is the bottleneck. Local SF data is hard to access, hard to work with: websites are slow, data sources are fragmented, and the tools they have to let people visualize trends often do not show visualizations with complete data, it's only a sample.
What we learned
911 calls related to guns are trending up.
What's next for Local Policy Summarizer
Expand this to measure 911 calls for all violent crime in SF. Let people add in the issues that they care about, and track what's being done about those issues
My presentation is here - https://docs.google.com/presentation/d/1rw1FoP_QUfzHdznDVe2sESC-orfEcoIa64WK4ePYmsk/edit#slide=id.p - please leave your thoughts and comments on how this might be improved, would be much appreciated.
Built With
- claude
- python
- sheets
Log in or sign up for Devpost to join the conversation.