Natural Language Processing with Python and spaCy: A Practical Introduction

by Yuli Vasiliev

Overview

An introduction to natural language processing with Python using spaCy, a leading Python natural language processing library.

Natural Language Processing with Python and spaCy will show you how to create NLP applications like chatbots, text-condensing scripts, and order-processing tools quickly and easily. You'll learn how to leverage the spaCy library to extract meaning from text intelligently; how to determine the relationships between words in a sentence (syntactic dependency parsing); identify nouns, verbs, and other parts of speech (part-of-speech tagging); and sort proper nouns into categories like people, organizations, and locations (named entity recognizing). You'll even learn how to transform statements into questions to keep a conversation going.

You'll also learn how to:

Work with word vectors to mathematically find words with similar meanings (Chapter 5)

Identify patterns within data using spaCy's built-in displaCy visualizer (Chapter 7)

Automatically extract keywords from user input and store them in a relational database (Chapter 9)

Deploy a chatbot app to interact with users over the internet (Chapter 11)

"Try This" sections in each chapter encourage you to practice what you've learned by expanding the book's example scripts to handle a wider range of inputs, add error handling, and build professional-quality applications.

By the end of the book, you'll be creating your own NLP applications with Python and spaCy.

Product Details

ISBN-13:	9781718500525
Publisher:	No Starch Press
Publication date:	05/12/2020
Pages:	216
Sales rank:	1,017,610
Product dimensions:	6.90(w) x 9.20(h) x 0.60(d)

About the Author

Yuli Vasiliev is a programmer, freelance writer, and consultant who specializes in open source development, Oracle database technologies, and natural language processing.

Table of Contents

Introduction xv

Using Python for Natural Language Processing xvi

The spaCy Library xvi

Who Should Read This Book? xvii

What's in the Book? xvii

1 How Natural Language Processing Works 1

How Can Computers Understand Language? 2

Mapping Words and Numbers with Word Embedding 2

Using Machine Learning for Natural Language Processing 3

Why Use Machine Learning for Natural Language Processing? 5

What Is a Statistical Model in NLP? 8

Neural Network Models 9

Convolutional Neural Networks for NLP 11

What Is Still on You 12

Keywords 12

Context 13

Meaning Transition 13

Summary 14

2 The Text-Processing Pipeline 15

Setting Up Your Working Environment 16

Installing Statistical Models for spaCy 16

Basic NLP Operations with spaCy 17

Tokenization 18

Lemmatizatton 18

Applying Lemmatization for Meaning Recognition 19

Part-of-Speech Tagging 21

Using Part-of-Speech Tags to Find Relevant Verbs 23

Context Is important 24

Syntactic Relations 24

Try This 28

Named Entity Recognition 29

Summary 29

3 Working with Container Objects and Customizing Spacy 31

SpaCy's Container Objects 31

Getting the Index of a Token in a Doc Object 32

Iterating over a Token's Syntactic Children 33

The doc.sents Container 33

The doc.noun_chunks Container 35

Try This 35

The Span Object 36

Try This 37

Customizing the Text-Processing Pipeline 37

Disabling Pipeline Components 38

Loading a Model Step by Step 38

Customizing the Pipeline Components 40

Using spaCy's C-Level Data Structures 42

How It Works 43

Preparing Your Working Environment and Getting Text Files 43

Your Cython Script 44

Building a Cython Module 45

Testing the Module 45

Summary 46

4 Extracting and Using Linguistic Features 47

Extracting and Generating Text with Part-of-Speech Tags 48

Numeric, Symbolic, and Punctuation Tags 48

Extracting Descriptions of Money 49

Try This 50

Turning Statements into Questions 51

Try This 54

Using Syntactic Dependency Labels in Text Processing 55

Distinguishing Subjects from Objects 55

Deciding What Question a Chatbot Should Ask 56

Try This 60

Summary 61

5 Working with Word Vectors 63

Understanding Word Vectors 64

Defining Meaning with Coordinates 64

Using Dimensions to Represent Meaning 65

The Similarity Method 66

Choosing Keywords for Semantic Similarity Calculations 67

Installing Word Vectors 68

Taking Advantage of Word Vectors That Come with spaCy Models 68

Using Third-Party Word Vectors 68

Comparing spaCy Objects 69

Using Semantic Similarity for Categorization Tasks 70

Extracting Nouns as a Preprocessing Step 71

Try This 72

Extracting and Comparing Named Entities 72

Summary 74

6 Finding Patterns and Walking Dependency Trees 75

Word Sequence Patterns 76

Finding Patterns Based on Linguistic Features 76

Try This 77

Checking an Utterance for a Pattern 77

Using spaCy's Matcher to Find Word Sequence Patterns 79

Applying Several Patterns 80

Creating Patterns Based on Customized Features 81

Choosing Which Patterns to Apply 83

Using Word Sequence Patterns in Chatbots to Generate Statements 83

Try This 86

Extracting Keywords from Syntactic Dependency Trees 86

Walking a Dependency Tree for Information Extraction 87

Iterating over the Heads of Tokens 87

Condensing a Text Using Dependency Trees 89

Try This 91

Using Context to Improve the Ticket-Booking Chatbot 91

Making a Smarter Chatbot by Finding Proper Modifiers 94

Summary 95

7 Visualizations 97

Getting Started with spaCy's Built-in Visualizers 98

DisplaCy Dependency Visualizer 98

DisplaCy Named Entity Visualizer 99

Visualizing from Within spaCy 100

Visualizing Dependency Parsing 100

Try This 102

Sentence-by-Sentence Visualizations 102

Customizing Your Visualizations with the Options Argument 103

Using Dependency Visualizer Options 103

Try This 104

Using Named Entity Visualizer Options 104

Exporting a Visualization to a File 106

Using displaCy to Manually Render Data 107

Formatting the Data 108

Try This 108

Summary 109

8 Intent Recognition 111

Extracting the Transitive Verb and Direct Object for Intent Recognition 112

Obtaining the Transitive Verb/Direct Object Pair 113

Extracting Multiple Intents with token.conjuncts 113

Try This 115

Using Word Lists to Extract the Intent 115

Finding the Meanings of Words Using Synonyms and Semantic Similarity 117

Recognizing Synonyms Using Predefined Lists 117

Try This 119

Recognizing Implied Intents Using Semantic Similarity 120

Try This 121

Extracting Intent from a Sequence of Sentences 122

Walking the Dependency Structures of a Discourse 122

Replacing Proforms with Their Antecedents 122

Try This 125

Summary 125

9 Storing User Input in a Database 127

Converting Unstructured Data into Structured Data 128

Extracting Data into Interchange Formats 129

Moving Application Logic to the Database 130

Building a Database-Powered Chatbot 132

Gathering the Data and Building a JSON Object 132

Converting Number Words to Numbers 133

Preparing Your Database Environment 135

Sending Data to the Underlying Database 137

When a User's Request Doesn't Contain Enough Information 138

Try This 139

Summary 139

10 Training Models 141

Training a Model's Pipeline Component 142

Training the Entity Recognizer 143

Deciding Whether You Need to Train the Entity Recognizer 143

Creating Training Examples 144

Automating the Example Creation Process 144

Disabling the Other Pipeline Components 146

The Training Process 146

Evaluating the Updated Recognizer 148

Creating a New Dependency Parser 149

Custom Syntactic Parsing to Understand User Input 149

Deciding on Types of Semantic Relations to Use 150

Creating Training Examples 150

Training the Parser 152

Testing Your Custom Parser 152

Try This 153

Summary 153

11 Deploying Your Own Chatbot 155

How Implementing and Deploying a Chatbot Works 156

Using Telegram as a Platform for Your Bot 157

Creating a Telegram Account and Authorizing Your Bot 157

Getting Started with the python-telegram-bot Library 158

Using the telegram.ext Objects 159

Creating a Telegram Chatbot That Uses spaCy 160

Expanding the Chatbot 161

Holding the State of the Current Chat 162

Putting All the Pieces Together 163

Try This 167

Summary 167

12 Implementing Web Data and Processing Images 169

How It Works 170

Making Your Bot Find Answers to Questions from Wikipedia 170

Determining What the Question Is About 171

Try This 174

Using Wikipedia to Answer User Questions 174

Try This 175

Reacting to Images Sent in a Chat 175

Generating Descriptive Tags for Images Using Clarifai 176

Using Tags to Generate Text Responses to Images 178

Putting All the Pieces Together in a Telegram Bot 178

Importing the Libraries 178

Writing the Helper Functions 178

Writing the Callback and main() Functions 180

Testing the Bot 181

Try This 182

Summary 183

Linguistic Primer 185

Dependency Grammars vs. Phrase Structure Grammars 185

Common Grammar Concepts 187

Transitive Verbs and Direct Objects 187

Prepositional Objects 187

Modal Auxiliary Verbs 188

Personal Pronouns 188

Index 189

Editorial Reviews

"A good resource for those programmers who want to learn to bridge the gap and write applications that anyone can use just by talking or writing to their machines and have the machine reply back."
—Jon Lazar, JustJon

From the Publisher