Being successful on twitter is difficult and a lot of users fail to make the most of the platform. We wanted to come up with a solution to test tweets before actually tweeting them und thereby boost their performance. Because we all know: in online marketing, performance is everything that matters.
What it does
Oracle scans all tweets from the last 7 days to analyze what made tweets successful. Based on parameters like hashtags, mentions, likes and retweets, we trained the machine to detect and develop patterns.
How we built it
We're using the twitter livestream API to come up with Hashtag suggestions. Periodically run word2vec to match the input text to possible hashtags - even while typing! With the RestAPI we can measure the potential success of the tweet.
Challenges we ran into
Implementing the twitter API without getting blocked, gather all the necessary data and create a solid prediction - all in less than 24 hours.
Accomplishments that we're proud of
It looks and feels great. We managed to create a great user experience that will enable even non-professionals to use our tool and optimize their twitter performance.
What we learned
Two days are a short period of time for this type of project. There is a future plan to improve the prediction massively by using the professional API GNIP from twitter.
What's next for Oracle
Developing an prediction engine, that's actually helpful.
Kristian Müller, Alexander Tonn, Daniel Seiler, Simon Karpstein, Martin Wiesemborski