madara (6) [Avatar] Offline
#1
I apologize in advance for my rough English!! smilie

I waited to read the whole book before give my opinion;

Well, sorry but this book disappointed my expectations..
the way to write, to explain is slightly inconsistent.. although the author's effort to be clear in the explanations is appreciable;`
many simple concepts are explained in a verbose way, uselessly;
many concepts are logically obvious and repeated too many times…like his code…(not optimized and not very efficient); and in any case: little code and too many words;

author boasts of teaching powerful tools for each kind of chatbot..
well, go to read!!

Generative model:

the seq2seq algorithm…( recovered from ‘Deep learning with Python’..as the author refers) but.. sorry in Deep learning with Python, exposition is more useful)… Furthermore you can exploit tools like https://github.com/farizrahman4u/seq2seq, based on keras…tutorial is more and more informative;

Retrieval-Based example:

Example really elementary, please…
just read some kernels on Kaggle’s Quora Question Pairs competition to learn more advanced and educational code IN ACTION , not only based on simple and standard cosine similiarity…

Search based chatbot:

just refer to good python framework like ChatterBot…no code, nothing news or interesting;

But The most absurd thing is:
pag. 420:

12.6 Four Wheel Drive.

"""As we promised at the beginning of this chapter, we’ll now show you how to combine all
four of these approaches to get traction with your users"""


but where??? please where author show that!???? NO ONE LINE OF CODE!! He only refers to Will!!..ehm...WILL!! .....forgive me but HERE author blur into the ridiculous: smilie))

Now, needed math was mentioned very superficially; clearly it is not the purpose of the book, but you can’t explain in educational way dimension reduction, backpropagation, activation function, etc.. those who do not have at least good basics in linear algebra and calculus…It does not make sense;

anyway chapters on CNN and RNN are explained quite clearly, appreciable..

In the end, this book is a good narration of what already exists, nothing news, no significant details, no topic is discussed in depth;

in entire book there ARE NOT examples of a chatbot in code!! (except for seq2seq generative TOY model), NO ONE!!

for me: NLP without ACTION
hobs (57) [Avatar] Offline
#2
I'm sorry NLPIA didn't meet your expectations, but I'm glad you found enough inspiration and resources to get you started.
You're right that the Chatbot chapter is short on end-to-end code examples, but I think you'll find the final version has a lot more detail to help you out:

* How to configure and customize open source chatbots like Chatterbot, Will, and AIMLBot
* How to generate completely new text using deep learning.

Some of the sentences in the book were composed by this generative model, but we couldn't train it until we had most of the manuscript completed. Do you think a sentiment analyzer will be able to tag the sentences of the book according to authorship: "Hobson", "Cole", "Hannes" or "Bot"?

We're working with other readers to build a chatbot that integrates all 4 of the chatbot approaches explained in chapter 12 and the block diagram that decorates the inside cover of the printed book. I don't know of any other open source package or online tutorial that even attempts to use all 4 chatbot approaches in a single package. I hope you'll consider contributing by submitting bug reports, feature requests, or pull requests at github.com/totalgood/nlpia.