We wanted to build a creative application where the computer could intelligently create a story about a given image. This could be used as a recreational app or an app for young students.
What it does
Provide an image. We'll identify the tags in the image, assess the attributes of this tag with respect to the English language, and use a simple grammar to generate a few sentences (assessing how valid each sentence is with 2-gram processing).
How we built it
We used a Django framework to wrap the site (the backend is in Python). We used clarif.ai and nltk to process data.
Challenges we ran into
We had some trouble with Django and connecting it to a domain with DigitalOcean. We also some trouble with processing huge college of n-grams (over a 1,000,000 line items). Natural language processing is also a challenge by itself.
What's next for Story Time
We further process the sentences, finding subjects and objects and relating them in future sentences. We need to optimize database transactions so the app is faster.