In today's day and age, there are an immense number of safety threats, especially with public events where large numbers of people attend. As college students, these safety concerns are amplified due to the rising threats of campus violence throughout the nation. We sought to combat this issue by creating a peer to peer alert system to ensure the safety of students and individuals.
What it does
Given an event, our application scours social media platforms such as Twitter to find related posts and performs natural language processing as well as computer vision techniques to construct a better context for the safety of an event. It also has a peer to peer alert system that allows students to forward safety risk information to other individuals in their geographical location
How we built it
We used React.js to develop the front-end and Flask to host our REST API. This API uses the TwitterAPI to find related tweets according to the given event. Then we used the Google Natural Language API to derive information from the text of each tweet as well as the Google Vision API to capture the expressions of faces in images associated with the post. Our user and safety data is stored on a Firebase Realtime Database that triggers alerts to users.
Challenges we ran into
We initially planned to use Facebook to gather information from a user's trusted network, but permission issues with the API prevented us from moving forward. Also, the Twitter API's documentation was convoluted and hard to follow, so a large amount of our time went to experimenting with the API
Accomplishments that we're proud of
Constructing a full stack application in 36 hours that leveraged APIs from several different contexts
What we learned
We learned how to partition tasks according to strengths more efficiently and how to work with large REST APIs
What's next for SentiGo
We plan to integrate more social media platforms and fine tune our algorithm for constructing safety data