Have you had “The Talk” with your chatbot about graph data structures?

A roundup of articles on improving chatbots with databases

Image credit: Charlotte Parent

This week our own Mark Watson was published in the Free Code Camp Medium publication. You should read it:

Have you had “The Talk” with your chatbot about graph data structures?

In his article, Mark describes how he modeled API interactions with the Watson Conversation service (no relation 😛) and the Spoonacular recipe API to work with the Apache Tinkerpop graph database framework. The result is a Slack chatbot that can use graph traversals to intelligently recommend recipes based on their popularity with other users.

The code and setup instructions are on GitHub in the watson-recipe-bot-nodejs-graph repo.

Hungry for more?

It builds off of two excellent articles, one by Josh Zheng and another from Mark.

Josh’s article reviews the important functions of the client application (in Python), and it walks readers through how to build the dialog flow in the Conversation Tool. It’s the place to start if you want to understand the inner-workings of the chatbot.

Mark’s previous article shows readers how to extend Josh’s chatbot by integrating the Cloudant JSON database service to cache API calls to Spoonacular. Aside from improved performance, Mark describes how to use the cached data to provide a more personalized user experience (for example, “show me my favorite recipes”). He also shows how to run some simple analytics on interactions, with an eye toward informing UX improvements. There are versions for Python and Node.js.

It was fun editing Mark’s work, especially after having recently completed some Coursera classes that emphasized graph data structures.

So check out Mark on Free Code Camp, and don’t forget to recommend the article to other intrepid Medium readers by clicking the ♡.

Thanks to Brad Noble and Quincy Larson for their feedback.