Build chatbot using Deep Natural Language Processing models in Tensorflow and Python
If you’ve ever used customer support livechat service, have you ever wonder with who are you actually chatting?
By 2020, Gartner estimates that chatbots will be handling 85 percent of customer-service interactions; they are already handling about 30 percent of transactions now. (Last update 11/2018)
- Python 3.5
- Tensorflow
- Numpy
To run the chatbot, run all code from the begining to Part 3 - Training the Seq2seq model. After that, you will obtained the training file 'chatbot_weights.ckpt'.
Training chatbot may take several days (+/- 5 days) depending on the power of the computer you are working on (CPU/GPU). It is also possible to use virtual machines (eg AWS), but this option is not free.
After that, to run the chatbot (eg in Spider terminal), run all code without "Part 3 - Training the Seq2seq model". Then you will be able to talk with your chatbot.