(this is one of the projects of the #openimpact shortlist)
The tool accesses the Twitter API to scan for instances of the phrase “food poisoning” tweeted within the geographic bounds of Chicago. Tweets caught by a classification algorithm are manually sorted for legitimacy and relevance, and any users identified as possible victims of food poisoning are tweeted a message to visit the Foodborne Chicago website, where they can report their illness to the CDPH via the city’s Open311 system. The health department then examines those cases the same as it does those received from all other channels.
|What does it do?||A twitter bot that searches tweets related to food poisoning in chicago to direct the victims to a 311-service form.|
|Outcome||Another channel to reach out to people affected by food poisoning, helping to prevent the spread of food borne diseases|
|Organizations||Smart Chicago Collaborative, Chicago Department of Public Health|
|Contact persons||Raed Mansour, Dr. Jason Miller, Daniel X. O’Neil|
|Media||youtube.com, govtech.com, cdc.gov (more here)|
|Source code||github.com (website), github.com (classifier), github.com (Tweet collector)|
|Data used||Twitter API|
|Data generated||Open 311 requests|
|Tech stack||Ruby on Rails, Node.js, MongoDB, R (some more info here)|
|Maintenance costs/month||Classification: 0.00001056 cents per Tweet, storage: 0.00000002 cents per Tweet per month (details see here)|
|First steps||1. Reach out to your local department of public health (or whoever organizes restaurant inspections)|
|2. Figure out how the Twitter bot can be integrated into their processes|
|3. Fork the code|
Interested in replicating this project? Join our DataLook #openimpact Slack channel or hit us up on Twitter.