Mucking Around with Watson Natural Language Understanding

Over the last couple of weeks I've been experimenting with Watson's APIs particularly those focused on analysis of text for sentiment, emotion, keywords and the like. I am particularly interested in how tools like Natural Language Understanding, Personality Insights and Tone Analyser can be leveraged to drive improved customer experience outcomes. IBM, Google and others are pushing these technologies into Contact Centres as you can see here however at CliqueAPI we're exploring how to leverage these tools in face to face scenarios as well.

In order to get a better feel for the tools and their power I've jumped back into doing some coding and created "Watson Starter". I am sharing Watson Starter here as I thought it may be beneficial to others that wanted to play around with the IBM APIs but found the existing demos a little too elaborate for their needs. I found the IBM samples had a lot of extra cruft thrown in. They seemed to focus more on graphic design than leveraging the APIs in meaningful ways. Good examples to show the capabilities but not great places to start for integrating into other projects.

Watson Starter is built using Express and NodeJS. Thus far I have only implemented basic processing and connectivity with the Natural Language Understanding API however I hope to add more shortly with the ultimate short term goal to capture a conversation either in a room or a CliqueAPI powered WebRTC call and deliver some insights back to the user.