Conversational AI: Chatbots that work

go Conversational AI brief contents contents preface acknowledgments about this book Who should read this book How this book is organized: A roadmap About the code liveBook discussion forum Other online resources about the author about the cover illustration Part 1: Foundations Chapter 1: Introduction to conversational AI 1.1 Introduction to AI assistants and their platforms 1.1.1 Types of AI assistants 1.1.2 A snapshot of AI assistant platforms 1.2 Primary use cases for AI assistant technology 1.2.1 Self-service assistant 1.2.2 Agent assist 1.2.3 Classification and routing 1.3 Follow along with this book 1.3.1 What you need to create your assistant 1.3.2 Useful spreadsheet software 1.3.3 Recommended programming language and code repository Chapter 2: Building your first conversational AI 2.1 Building a conversational AI for Fictitious Inc. 2.2 What’s the user’s intent? 2.2.1 What’s an utterance? 2.2.2 What’s a response? 2.2.3 How does the assistant understand what the user means? 2.2.4 Why machine learning? 2.2.5 What’s an intent? 2.2.6 What’s an entity? 2.2.7 Combining intents and entities 2.2.8 Adding intents to the Fictitious Inc. assistant 2.3 Responding to the user 2.3.1 Simple question-and-answer responses 2.3.2 Contextualizing a response by using entities 2.3.3 An alternative way to provide contextual responses 2.3.4 Responding with a process flow 2.4 Other useful responses 2.4.1 Detecting low confidence 2.4.2 Counting misunderstandings 2.4.3 Implementing confidence detection and the two-strikes rule 2.5 Try to build this conversational AI yourself Part 2: Designing for success Chapter 3: Designing effective processes 3.1 What processes will the assistant handle? 3.1.1 Designing for the most common user needs 3.1.2 Assembling a plan and a dream team 3.1.3 Managing the design process 3.1.4 Cross-cutting design aspects 3.2 Choosing the channel to implement first 3.2.1 How users receive information in voice and web 3.2.2 How the assistant receives information in voice and web Chapter 4: Designing effective dialogue 4.1 Writing dialogue 4.1.1 Take a conversational tone 4.1.2 Don’t repeat yourself (much) 4.1.3 Acknowledge the user 4.2 Asking questions 4.3 What if the assistant doesn’t understand? 4.3.1 Reprompting 4.3.2 Disambiguation 4.3.3 Escalation Chapter 5: Building a successful AI assistant 5.1 AI assistant use cases 5.2 Conversational AI success metrics 5.2.1 Containment 5.2.2 Time to resolution 5.2.3 Net promoter score 5.2.4 Coverage 5.2.5 Instrumenting your conversational AI 5.3 Command interpreter success metrics 5.3.1 Usage 5.3.2 Stickiness 5.3.3 Instrumenting your command interpreter 5.4 Event classifier success metrics 5.4.1 Time to resolution 5.4.2 Number of hand-offs 5.4.3 Other customer satisfaction metrics 5.4.4 Instrumenting your classifier Part 3: Training and testing Chapter 6: Training your assistant 6.1 Training an AI assistant 6.2 Finding training data 6.2.1 Production logs 6.2.2 A mock user interface 6.2.3 Subject-matter experts 6.2.4 Organizing training data into intents 6.3 Assessing whether you have the right training data 6.3.1 Training data variety 6.3.2 Training data volume 6.3.3 Training data veracity Chapter 7: How accurate is your assistant? 7.1 Testing an AI assistant for accuracy 7.1.1 Testing a single utterance 7.1.2 Testing multiple utterances 7.1.3 Selecting a test data set 7.2 Comparing testing methodologies 7.2.1 Blind testing 7.2.2 k-folds cross-validation test 7.3 Selecting the right accuracy metric for the job Chapter 8: Testing your dialogue flows 8.1 Functionally testing a dialogue flow 8.1.1 Manually testing a conversation flow 8.1.2 Automating a conversation flow test 8.1.3 Testing the dialogue flowchart 8.1.4 Testing the unexpected error paths 8.2 Nonfunctionally testing a dialogue flow 8.2.1 User experience testing 8.2.2 Load testing Part 4: Maintenance Chapter 9: Deployment and management 9.1 Where to store your code 9.1.1 Taking the Wild West approach 9.1.2 Using source control for code 9.2 Where to run your code 9.2.1 Development environment 9.2.2 Test environment 9.2.3 Production environment 9.2.4 After the first production deployment 9.3 Using source control for other assets Chapter 10: Improving your assistant 10.1 Using a success metric to determine where to start improvements 10.1.1 Improving the first flow to fix containment problems 10.1.2 Inspecting other process flows for containment problems 10.2 Analyzing the classifier to predict future containment problems 10.2.1 Representative baseline 10.2.2 Finding gaps in the training data 10.3 When and why to improve your assistant 10.3.1 You can’t fix everything at once 10.3.2 You can’t always predict how users will react 10.3.3 User needs will change 10.3.4 Not every problem is technical Part 5: Advanced/optional topics Chapter 11: Building your own classifier 11.1 Why build your own classifier? 11.1.1 Classification is a differentiator 11.1.2 Classification is a core competency 11.1.3 Traceability 11.1.4 To learn 11.1.5 Build or buy? 11.2 Build a simple classifier from first principles 11.2.1 The simplest text classifier 11.2.2 The mathematics behind a simple classifier 11.3 Expanding the simple classifier 11.3.1 Predicting more than one class 11.3.2 An all-in-one classifier 11.3.3 Comparing binary classifiers to all-in-one classifiers 11.4 Extending even further 11.4.1 What happens when you add more training data? 11.4.2 Exercise: Experiment on your own Chapter 12: Additional training for voice assistants 12.1 Collecting data to test a speech-to-text model 12.1.1 Call recordings as speech training data 12.1.2 Generating synthetic speech data 12.2 Testing the speech-to-text model 12.2.1 Word error rate 12.2.2 Intent error rate 12.2.3 Sentence error rate 12.3 Training a speech-to-text model 12.3.1 Custom training with a language model 12.3.2 Custom training with an acoustic model 12.3.3 Custom training with grammars appendix: Glossary of terms and abbreviations index A B C D E F G H I J K L M N O P Q R S T U V W