The popularity of U of T Confessions among U of T students and how the lack of a good interface made access to very large percentages of wanted data being practically inaccessible to people. We realized that other university confession pages also shared a similar problem and expanded our scope for all universities.
What it does
Our website provides the ability for users to query through the endless data offered in picture form by university Instagram confession pages, particularly U of T's page "uoftears_". It allows users to sort the nearly 4000 confessions on the page by any metric they may choose. Some examples are: most liked, most comments, most related to a certain topic, most controversial, most popular, most recent, and many other metrics.
How we built it
It was built by first extracting each of the pictures and their priceless metadata from the Instagram page. The metadata contained the number of people who like the post, number who commented, specifically who commented, what they commented, number of likes per comment, date and time, and captions. Additionally we gathered even more information by using the Google cloud vision and natural language processing APIs on the pictures themselves to extra post number, confession text, and year and faculty of poster. Using all this data we set up a powerful database and backend that continuous updates it, plus a front end to take user queries.
Challenges we ran into
- Setting up the front end
- Data mining and organizing this vasts amounts of data
Accomplishments that we're proud of
- An incredible database of huge amounts of metadata associated with each image that is continuously updated
What we learned
- Web dev
- Back end
- When not to use Blockchain
What's next for UniConfessions
- To take over the university confessions industry