Python Chat Bot Tutorial - Chatbot with Deep Learning (Part 2)

Welcome to part 2 of creating a chatbot with deep learning and python. In this part of the series we continue to preprocess our data into whats known as a bag of words! This will allow us to feed numerical data to our model in the next part.

◾◾◾◾◾
💻 Enroll in The Fundamentals of Programming w/ Python

◾◾◾◾◾◾

⚡ Please leave a LIKE and SUBSCRIBE for more content! ⚡

Tags:
- Tech With Tim
- Python Chat bot
- Chat bot tutorial python
- Python chat bot tutorial
- Deeplearning chatbot python
- Python Tutorials

[ngapis]

line 37: out_empty = [0 for _ in range(len(classes))] #should be: out_empty = [0 for _ in range(len(labels))]
line 57: output = numpy.array(output)
#or change import numpy as np
line 56: trainig = np.array(trainig)
line 57: output = np.array(output)

[khaleda.s]

Tim isn't great at explaining, but I really appreciate his tutorials since they're straight to the point

[bfkmnemonic]

Can't wait for it to be done.. I really need someone to talk to!

[dr.mikeybee]

Good job. Some of the code is very dense; so that makes it a bit more difficult to learn the important topics like bag of word. If you do some of the single line statements as multiline loops, rather than enumerations, it would make things more easily understandable for more levels of programmers. It's hard to teach a lot of people if you don't simplify where you can. Nevertheless, I appreciate your direct style. You get to the point quickly.

[josueal.]

Has anyone considered using a python dictionary to store our "docs_x" and "docs_y"? A dictionary feels like a more intuitive data structure. In essence, we can search the patterns using the tags as keys.

[everything]

Great. You make the things clear very well.

[nikhilambavaram]

Thank you, Tim for making these awesome videos! I have learned so much!

[naveenagrawal_nice]

hey tim, it is asked by another user but you haven't answered it so i am asking again
please reply asap
for x, doc in enumerate(docs_x):

bag = []

wrds = [stemmer.stem(w.lower()) for w in doc]


isn't that part wrong?
We get a sentence for doc in each loop. Then we put doc in a loop. Problem occurs here.
w gets letter in each loop and stem of letter is the same as letter. But we want w gets a word and wrds will be stem of words.

[funyn]

so is an encoded bag of words kind of like a hash in which different patterns will always output a different encoded bag of words

[TRIPDoesItAll]

hi, amazing tutorial.i have one question, if i wanted my bot to be able to recognize synonyms using wordnet's sysnsets, how would i implement that?

[channeljukebox]

Amazing tutorial. Thanks bro. Can you make some tutorial on RASA as well? Thanks in advance.

[deepakrathi]

Resource punkt not found.
Please use the NLTK Downloader to obtain the resource:
what's this error?

[nstcg]

stem() missing 1 required positional argument: 'word'

did anyone had the same error?

[hemanko]

on line 37 what is classes where did he declare it?

[Tom-kgqe]

Your word classifier is very similar to the Pandas.Get_Dummies function, except I'm not sure that it would take into account the labels part...

[secretr]

I get an error with tensor flow saying cudart is not found My version of tensorflow is 2.6.2 can you help me?

[eugenechua]

just a quick question..i still dont quite get the part when he did what is the doc_y[x] trying to accomplish? converting the tags to 1?

[masonqian]

Hey Tim, great series. I had a question about the enumeration loop specifically. Does it matter that duplicates aren't being addressed? Within a pattern the same word might be seen more than once ie: "THE weather today is THE hottest it has been", and the conditional "if w in wrds" would evaluate twice and append twice. Also is the one hot encoding model for a bag of words appropriate for a more complex chatbot? How much data does it handle well?


Edit: I think it might have implications in this line as well: = 1" since .index() will only return the index of the first occurrence.

[haswanthaekula]

Why should not we shuffle our data? I did not get that part. It would be great if you can explain that. Thanks

[basketballgirl]

What is the difference between the words list and wrds? Seems like both are list of words from pattern. If so how does the for w in words if w in wrds: to get the 0 and 1 work? Wouldn’t all be one because they are the same list. Super basic question would really appreciate a response!!! Mine is working but I don’t get the logic :(