Ask the Critics

Movie reviews curated by New York Times critics are an utterance away. Ask for the critics picks, get their reviews.

Comment

Inspiration

It's nearing the end of my second year in college, Q's third, and we both came to the realization: we haven't really created anything. Inspired by the plethora of technologies to fiddle and tinker with, we came to an idea of having an Amazon Echo tell you things, considering that is pretty neat.

We thought that using a dataset for movies and then training a recommendation algorithm would be pretty cool. There were a ton of recommendation engines to learn from, yet, considering the depth of those topics and this being our first hackathon, we ended up scaling back ambitions appropriately.

What it does

Utilizing Python's Flask, New York Times' movie data set, and MongoDB, we were able to run a local web service that the provided Amazon Echo could communicate with and interact with appropriately. Based off of voice queries processed by the Amazon Echo, we could control responses and data that comes in, and return appropriate responses in return.

How we built it

Google, Python, Google, MongoDB, Google, more API documents, Google, StackOverflow, Google...

Challenges we ran into

  • Realizing that a recommendation system was hard.
  • Realizing that none of these recommendation systems can query all the data reasonably well.
  • Finicky voice recognition.
  • Being unable to recognize a new line prevented proper database connections for the longest of time.

Accomplishments that we're proud of

  • Good harmony amongst data sets in order to present data.
  • Being able to make the Echo speak out words that we're controlling.
  • Being able to craft language to make the Echo output onto the controlling device.

What we learned

  • Double-checking dashes and letters is really important.
  • Creating something akin to global variables from the get-go so that no one accidentally commits a key.
  • Sleep is a mere metaphor that's not necessary.

What's next for Ask the Critics

  • Probably a good, long nap.
  • Could be extended to being able to look up a result from the MongoDB instance, and if it's not there, query NYTimes, cache the results, and then serve them up. If only Flask-Ask supported rich media in content responses...
Share this project:

Updates