Clients were having a difficult time making sense of the true power of natural language processing and how it could be leveraged to better understand their customers in the social sphere. Two short weeks later, we built Trump Master Flex to help bring our capabilities to life.
We were limited in time and internal resources, and of course all of the regulations that come with Twitter, Spotify, and IBM Watson. Additionally, with only 280 characters to pull data from, our source content was limited. Therefore, we designed our custom-built machine learning model to make contextual word associations, allowing us to expand our data pool. We also had to add a filter to black list certain words (i.e. Terror, Police, etc.) so that we did not end up making a pairing that came off as insensitive to the context of the tweet.