The inspiration from this comes from a constant desire to find new anime shows and the lack of great recommendation systems other than word of mouth. So the idea was to create an app that has an interface that is familiar to many people in order to build a better recommendation system that incorporates as much data as possible into it.
What it does
This app creates cards just like tinder and allows users to swipe left or right on the to either ignore the shows or add them to their plan-to-watch queue. This queue then interfaces with the users myanimelist account and can update their profile to reflect that they wish to watch the shows that they swipe right on.
How we built it
We built it using Ionic, Angular, MongoDB, and NodeJS to built a flexible but easy to use system to create our app on top of. We scraped image files and summaries for every show on myanimelist. We already had the list of all shows, their ratings, genres, and how many people have rated the show (acquired from kaggle). We added all the data to MongoDB and focus on using RESTful API to interface with the data. The tinder template comes from ionic and was integral in getting our app up and running as quickly as we could with only 3 group member. With ionic we are able to deploy the app to phones and show off its true potential.
Challenges we ran into
We ran into challenges with making HTTP GET requests in angular as there was not a large amount of documentation and many places to mess things up slightly, we ended up spending 6 hours trying to get this working. We ran into a little issue with git that took an hour to fix and the scraping of data from myanimelist was slow. SOme of the data we scraped still has some html tags in it and we did not have time to remove them.
Accomplishments that we're proud of
We have a pretty app that is able to combine Summary, Cover Photo, and Show data and allows users to swipe right and left on it. We are able to interface with myanimelist to update users show list and verify that they are indeed members of myanimelist. We have a user collection that is able to store users. **We had a lot of fun!!!**
What we learned
We learned that ionic may not have been the best for building the app and we would have gotten a lot more use out of a python backend that we could build a much more robust reccomendation system off of and incorporate their libraries. We learned that even a seemingly small project such as ours still could have used more members.
What's next for Tinder for Anime
Next we will finish the reccomendation system and the interface when users swipe right to show the shows they have swiped right on in a pretty list. Then after the recommendation system is fleshed out we would like to get it hosted and out there for people to use.