Inspiration
Every once in a while, we want to watch a movie that we haven't seen before, but we don't know which one. There is lots of content on Netflix to consume, but are those really worth watching? Who will know what types of movies you like? We are united by a desire to create a new experience in movie selection, a more personalized and experimental approach than any previous engine. Meet with ORDA, your new suggestion friend!
What it does
Semantic Movie Finder
The input can be a couple of sentences, but a paragraph too. Using that information, we will query our big database of 40000+ movies and find the best possible movies for you. Our system will make a semantic search and find you the best matches. Just go to the website and prompt what you want to watch!
Orda Tinder
If you don't know what to write, that is fine, we can create the options by showing you the samples, and it creates a profile to find the best suitable movies for you! By looking at the movie posters, click Like or Dislike, so that we will have a notion about your movie aesthetics. We will combine that information to suggest you a new movie!
How we built it
Using state-of-the-art text encoders, we can convert your input into something machine-friendly. After converting it into a vector representation, we search through our database to find the closest samples. Since there are more than 40.000 movies, we used a quantization algorithm to speed up the search process. It won't take a second, we promise!
Challenges we ran into
Aggregating tons of data was quite a thing. There are different data sources out there, we need to clean that in great detail to make it work. Another issue was API usage since most of the movie APIs were paid. We ended up paying a small amount to use an API and save time.
Accomplishments that we're proud of
Our system can generate quite interesting movies, that are worth watching and it can do it in seconds. The overall system is well-designed and has everything in its right place. The front end is easily extendable with React components. New recommendation algorithms can be implemented very easily with just a simple API endpoint.
What we learned
We learned teamwork can beat any problem. Our front-end people aced react.js. Our business management person practiced python. Our machine learning people learned front-end, and we were able to do it in a matter of hours. As a team, we learned how valuable we are.
What's next for Orda
We plan to launch our application for casual use!
Log in or sign up for Devpost to join the conversation.