Summary

In this chapter, we referred to a different dataset in order to make a chatbot. You learned about the rule-based approach that can be used if you don't have any datasets. You also learned about the open and closed domains. After that, we used the retrieval-based approach in order to build the basic version of a chatbot. In the revised approach, we used TensorFlow. This revised approach is great for us because it saves time compared to the basic approach. We implemented Google's neural Conversational Model paper on the Cornell Movie-Dialogs dataset. For the best approach, we built a model that used the Facebook bAbI dataset and built the basic reasoning functionality that helped us generate good results for our chatbot. Although the training ...

Get Machine Learning Solutions now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.