Before the keynote, while the hackers were busy stewing in their own juices, heating slowly by proximity to crowd and boredom, we were treated to a presentation of all the tweets that went @dubhacks, or contained the #dubhacks. Unfortunately for this brilliant time wasting idea it quickly devolved into dankness, both in terms of memes and otherwise. By the time cocaine came around, the plug had to be pulled.
We thought this was a shame. We thought if only freedom of speech was constricted in some way, censored in fact, the screen could show, nice, vanilla,
state sponsored, images and tweets with little to no moderator effort.
What it does
This app takes images with a certain hashtag, runs them through Clarifai's API for tags, and censors any image with one or more undesired tag (as specified by the user)
How I built it
Challenges I ran into
AWS proved difficult to use in our sleep-deprived state, especially since none of our group had much if any experience in web hosting. While we got through it in the end, we spent more than three hours debugging a single, now dreaded, error message. We also had a lot of trouble using the part of Clarifai's API for training new tags, to the point we were forced to abandon that part of our project.
Accomplishments that I'm proud of
I'm very happy that the project managed to come together, despite all of us having to pull together to learn multiple new ideas (some of our group was writing Python for the first time), and despite none of us knowing each other before the competition.
What I learned
I personally learned a lot about web hosting, and simple http requests. My group members report much of the same, along with learning in Python, and even in use of Github.
What's next for Twitter Filter
We aren't sure what will come next for this app idea. We have lots of ways to flesh it out, especially involving analysis of tweet text itself, and use of Clarifai's training algorithms to allow users to create custom tag bans.