I originally posted a few useful Watson Conversation articles and videos on my blog, but they weren’t at all organised, and there are new articles all the time.
To make it easier to get started with the Conversation service, and hopefully to keep up with new articles and changes to the service, I’m planning to create a quick start guide.
Please open an issue on GitHub if you spot any problems, or know of any resources that it would be worth including!
Before diving in and getting started with the Conversation service, it’s worth taking a bit of time thinking about what you are trying to achieve. Watson Conversation can be used instead of the Natural Language Classifier service for a variety of purposes but the chances are you’re thinking about creating a chatbot.
For some clues to why chatbots are popular, and a few likely uses, take a look at Simon Burns’ Seriously, what do I need a bot for? article. I’m also on the look out for other examples of what’s possible, for example Sayfe Space from the Disrupt London 2016 Hackathon so please get in touch if you know of any!
3 steps for building a cognitive solution in 70 days is also worth a read to get an idea of the full process, and the work required to prepare your data and train the system.
Even when you’ve decided that a chatbot is the way to go, it’s still worth considering whether it should chat or not! Does Conversation Hurt Or Help The Chatbot UX? explores two different perspectives.
Bluemix is IBM’s cloud offering, which you can also use to bring your cognitive applications to life. If you’re new to Bluemix or Watson, the Watson Developer Cloud Getting Started documentation is a good place to begin.
Don’t worry if you don’t want to use Bluemix though; once you’ve created a Conversation service you can call it from anywhere.
Likely/possible next steps:
- Creating a Conversation service
- Entities (including turning on all the system entities!)
- Linking to Slack/Facebook/Friendface
- Integrating with other services
- Finding help
Plus an about page, other resources and examples. Comments, suggestions, links and pull requests always welcome!