The inception of this project comes from a hackfest together with a partner; SocialBoards. They offer a unified support inbox aggregated from different sources, such as social media. To ease with tagging and understanding of the messages coming in, with things like sentiment analysis, the customers of SocialBoards will get an improved and more optimal workflow.
Machine Learning and specifically Cognitive Services are really popular these days. The possibilities seem to be endless. Microsoft offers cognitive services and within it you'll find the text analytics API. Something SocialBoards had been looking into and hoping to utilize. Unfortunately, the classifications, extracting key phrases and sentiment analysis does not as of May 2017 support Norwegian. Together with the partner we did an analysis of the needs and decided on a time and place for a 2 day hackfest to do a proof of concept. Within the 2 days we managed to get the project off the ground and prove that it was doable with out of the box NodeJS packages.
Below is the the initial design sketch for what we wanted to build, including technologies and some REST APIs we early on defined:
