What do you think?
Rate this book
Natural Language Processing in Action
Natural Language Processing in Action is your guide to creating machines that understand human language using the power of Python with its ecosystem of packages dedicated to NLP and AI! You'll start with a mental model of how a computer learns to read and interpret language. Then, you'll discover how to train a Python-based NLP machine to recognize patterns and extract information from text. As you explore the carefully-chosen examples, you'll expand your machine's knowledge and apply it to a range of challenges, from building a search engine that can find documents based on their meaning rather than merely keywords, to training a chatbot that uses deep learning to answer questions and participate in a conversation.
544 pages, Paperback
Published April 14, 2019
47 people are currently reading
246 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 11 of 11 reviews
October 30, 2018
An extremely good read on NLP / ML / DL topics. Mostly focused on DL for natural language processing/understanding/generation.
June 16, 2021
This book is somewhere between a recipe book and a textbook. It's got real examples of various NLP techniques, like a recipe book, but there are not enough of them to make it one; it's got some theory and history, like a textbook, but not rigorously or systematically enough to make it one. I read it as an introduction to NLP practice, and for that it's very useful (not least because it was published only two years ago as of this writing).
Explanations are usually precise and make sense, although the at the points where the prose is less than clear I'm not sure whether it's just a consequence of the difficult subject matter. The authors have adopted a casual style, similar to that of MOOC professors. While this, again, usually works, it sometimes feels like they're pandering or affecting that language. Admirably, the authors have generally chosen to highlight modern techniques to solving NLP problems while contextualizing them in a bit of history. They don't often fall into the textbook trap of discussing outmoded algorithms with such enthusiasm that you can't help but suspect that one of the authors did their PhD on it twenty years ago. (That said, I could have done without the extended discussion of the history of chatbots and AIML, which seems to have been one of those "standards" dreamed up in the academy and never actually used.)
Speaking of chatbots. the main conceit of the book is that they'll teach you all the building blocks on the road to making a chatbot. (Coincidentally — or not — one of the authors worked at Aira, a company that built a chatbot for the visually impaired.) This conceit sometimes becomes strained (they often say, "Here's a technique you might use in some other application, but let's talk about this other algorithm, which you'll actually use for the chatbot," and then they don't actually use it). And they don't even really teach you how to make a chatbot. But this is okay, because their explanations of the techniques are good enough. There is a little too much currency given to classical techniques — if-then trees, regular expressions, etcetera. I understand their importance in the past, and perhaps even today in some applications, but now NLP means almost solely AI techniques. They know this too, and don't develop the classical ideas enough for a reader to understand why in the world someone would use them rather than AI.
In the preface, the authors, who are three white guys, claim that the "outsized impact" of "prosocial, cooperative chatbots… is why and how the authors of this book came together… We hope that our words will leave their impression in your mind and propagate like a meme through the world of chatbots, infecting others with passion for building prosocial NLP systems." I suspect Hobson wrote this sentence, since in his bio he wrote that he's "passionate about openness and prosocial AI." Yet the book hardly touches on the ethical implications of their actions, giving only a few brief asides about some historical gaffes in production NLP systems. They tell you that biased data gives biased results, which is common sense; they don't tell you /how/ to get unbiased data or how to deal with the biased data you've got. It often seems that on the rare occasions when they do speak of prosocial AI, they think that "prosocial" means "kind" rather than "just" or even "good." While they're probably trying their best to address the ethics of AI, you can tell that they're engineers first.
I think that the greatest strength of the book is its clarity and its breadth. I have only a working knowledge of Python and Keras and linear algebra, and yet I was able to understand most of what they were saying (even when they were discussing the algorithms behind the Keras one-liners). And although the book is far from comprehensive, I feel that I have a decent foothold in field from which to branch out to deeper understanding. The authors also provide a Python library to go along with the book, which I haven't had the occasion to try, but it seems like they put a lot of effort into it.
Overall, the book is well-written and well-explained. It suffers from a rather loose structure and a lack of commitment to the theme (prosocial chatbots), but neither of these are grave faults. 3.7 stars.
Explanations are usually precise and make sense, although the at the points where the prose is less than clear I'm not sure whether it's just a consequence of the difficult subject matter. The authors have adopted a casual style, similar to that of MOOC professors. While this, again, usually works, it sometimes feels like they're pandering or affecting that language. Admirably, the authors have generally chosen to highlight modern techniques to solving NLP problems while contextualizing them in a bit of history. They don't often fall into the textbook trap of discussing outmoded algorithms with such enthusiasm that you can't help but suspect that one of the authors did their PhD on it twenty years ago. (That said, I could have done without the extended discussion of the history of chatbots and AIML, which seems to have been one of those "standards" dreamed up in the academy and never actually used.)
Speaking of chatbots. the main conceit of the book is that they'll teach you all the building blocks on the road to making a chatbot. (Coincidentally — or not — one of the authors worked at Aira, a company that built a chatbot for the visually impaired.) This conceit sometimes becomes strained (they often say, "Here's a technique you might use in some other application, but let's talk about this other algorithm, which you'll actually use for the chatbot," and then they don't actually use it). And they don't even really teach you how to make a chatbot. But this is okay, because their explanations of the techniques are good enough. There is a little too much currency given to classical techniques — if-then trees, regular expressions, etcetera. I understand their importance in the past, and perhaps even today in some applications, but now NLP means almost solely AI techniques. They know this too, and don't develop the classical ideas enough for a reader to understand why in the world someone would use them rather than AI.
In the preface, the authors, who are three white guys, claim that the "outsized impact" of "prosocial, cooperative chatbots… is why and how the authors of this book came together… We hope that our words will leave their impression in your mind and propagate like a meme through the world of chatbots, infecting others with passion for building prosocial NLP systems." I suspect Hobson wrote this sentence, since in his bio he wrote that he's "passionate about openness and prosocial AI." Yet the book hardly touches on the ethical implications of their actions, giving only a few brief asides about some historical gaffes in production NLP systems. They tell you that biased data gives biased results, which is common sense; they don't tell you /how/ to get unbiased data or how to deal with the biased data you've got. It often seems that on the rare occasions when they do speak of prosocial AI, they think that "prosocial" means "kind" rather than "just" or even "good." While they're probably trying their best to address the ethics of AI, you can tell that they're engineers first.
I think that the greatest strength of the book is its clarity and its breadth. I have only a working knowledge of Python and Keras and linear algebra, and yet I was able to understand most of what they were saying (even when they were discussing the algorithms behind the Keras one-liners). And although the book is far from comprehensive, I feel that I have a decent foothold in field from which to branch out to deeper understanding. The authors also provide a Python library to go along with the book, which I haven't had the occasion to try, but it seems like they put a lot of effort into it.
Overall, the book is well-written and well-explained. It suffers from a rather loose structure and a lack of commitment to the theme (prosocial chatbots), but neither of these are grave faults. 3.7 stars.
July 12, 2021
After one of my master's projects, I had realized how little I knew with NLP. Going through the book I got a lot of ideas on how people actually use text. I highly recommend this book for anyone interested in the subject.
January 7, 2020
Too wordy
January 2, 2022
You really gotta walk a long way before before you get anywhere with NLP.
July 31, 2021
Excellent introduction for those who have some background in NLP and would like to learn something about the state-of-the-art in NLP with TensorFlow.
I strongly recommend to anyone interested. The book is condensed with examples with very intuitive explanations.
I strongly recommend to anyone interested. The book is condensed with examples with very intuitive explanations.
July 6, 2024
A practical introduction to the topic of natural language processing with many examples in Python. The book explains the important concepts in an understandable way and without the need to sound clever.
April 23, 2020
It cover quite a few with good code sample. Recommended!
January 1, 2024
A clear and informative book about basic of NLP. I can easily recommend it to new engineers in this field of machine learning
March 21, 2024
I want a textbook, not a cookbook.
April 2, 2020
If there's one book worth its weight in gold, this would be it.
I'm impressed by the thorough analysis made on the fascinating field of Natural Language Processing made by the authors. They present the latest techniques AND code examples. Most books I've read present the mathematics but almost no reproducible code so I find it very difficult to test and visualize what they're explaining.
Not so with this text. Definitely need to reread again. I struggled mostly because of how much data it presents, that makes it hard to simply skim. Definitely need to re-read and retry the exercises. If any weakness I could find is only in latest chapters where they go in depth on the latest techniques, that they present Keras and Finite State Machines which are presented in Python code but not explained as that would make the text even bigger than it already is.
I'm impressed by the thorough analysis made on the fascinating field of Natural Language Processing made by the authors. They present the latest techniques AND code examples. Most books I've read present the mathematics but almost no reproducible code so I find it very difficult to test and visualize what they're explaining.
Not so with this text. Definitely need to reread again. I struggled mostly because of how much data it presents, that makes it hard to simply skim. Definitely need to re-read and retry the exercises. If any weakness I could find is only in latest chapters where they go in depth on the latest techniques, that they present Keras and Finite State Machines which are presented in Python code but not explained as that would make the text even bigger than it already is.
Displaying 1 - 11 of 11 reviews