Artificial Intelligence for Customer Relationship Management: Solving Customer Problems

✍ Scribed by Boris A. Galitsky

• Publisher: Springer
• Year: 2021
• Tongue: English
• Leaves: 474
• Series: Human–Computer Interaction Series
• Category: Library

✦ Synopsis

The second volume of this research monograph describes a number of applications of Artificial Intelligence in the field of Customer Relationship Management with the focus of solving customer problems. We design a system that tries to understand the customer complaint, his mood, and what can be done to resolve an issue with the product or service. To solve a customer problem efficiently, we maintain a dialogue with the customer so that the problem can be clarified and multiple ways to fix it can be sought. We introduce dialogue management based on discourse analysis: a systematic linguistic way to handle the thought process of the author of the content to be delivered. We analyze user sentiments and personal traits to tailor dialogue management to individual customers. We also design a number of dialogue scenarios for CRM with replies following certain patterns and propose virtual and social dialogues for various modalities of communication with a customer. After we learn to detect fake content, deception and hypocrisy, we examine the domain of customer complaints. We simulate mental states, attitudes and emotions of a complainant and try to predict his behavior. Having suggested graph-based formal representations of complaint scenarios, we machine-learn them to identify the best action the customer support organization can chose to retain the complainant as a customer.

✦ Table of Contents

Preface to Volume 2
Acknowledgements
Contents
1 Chatbots for CRM and Dialogue Management
1.1 Introduction: Maintaining Cohesive Session Flow
1.1.1 Current State-of-the-Art: Not Good for CRM
1.2 Chatbot Architectures and Dialogue Manager
1.3 Building Dialogue Structure from a Discourse Tree of an Initial Question
1.3.1 Setting a Dialogue Style and Structure by a Query
1.3.2 Building a Dialogue Structure in Customer Support Dialogues
1.3.3 Finding a Sequence of Answers to be in Agreement with a Question
1.3.4 Searching for Answers with Specified RR for Dialogue Construction
1.3.5 Datasets for Evaluation
1.3.6 Evaluation of the Dialogue Construction from the First Query
1.4 Dialogue Management Based on Real and Imaginary Discourse Trees
1.4.1 Answering Questions via Entities and Discourse Trees
1.4.2 Question Answer Filtering Algorithm
1.4.3 Experiments with Answering Convergent Questions
1.5 Dialogue Management Based on Lattice Walking
1.5.1 Formal Concept Analysis
1.5.2 Lattice Walker Example
1.5.3 Lattice Navigation Algorithm
1.6 Automated Building a Dialogue from an Arbitrary Document
1.6.1 Forming a Dialogue from a Discourse Tree of a Text
1.6.2 Question Formation and Diversification
1.6.3 System Architecture
1.6.4 Evaluation of the Dialogue Builder
1.6.5 Rhetorical Map of a Dialogue
1.6.6 Evaluation of Chatbot Performance Improvement Building Datasets on Demand
1.7 Open Source Implementation
1.8 Related Work
1.9 Conclusions
1.9.1 Conclusions on Building a Dialogue
References
2 Recommendation by Joining a Human Conversation
2.1 Introduction
2.2 Slot-Filling Conversational Recommendation Systems
2.3 Computing Recommendation for a Dialogue
2.4 Assuring the Recommendation is Persuasive and Properly Argued For
2.5 Continuing Conversation with RJC Agent
2.6 System Architecture
2.7 Evaluation
2.8 Related Work and Discussion
References
3 Adjusting Chatbot Conversation to User Personality and Mood
3.1 Introduction
3.2 Recognizing Personality
3.3 Models of Emotions
3.4 Transitions Between Emotions
3.5 Emotion Recognition Datasets
3.6 Emotional Selection System Architecture
3.7 Evaluation
3.8 Emotional State Transition Diagram in CRM
3.9 Related Work and Conclusions
References
4 A Virtual Social Promotion Chatbot with Persuasion and Rhetorical Coordination
4.1 Introduction
4.2 Conducting Virtual Dialogue
4.2.1 A Session with Virtual Dialogue
4.2.2 A Persuasive Dialogue
4.2.3 Dialogue Management
4.2.4 Dialogue Construction from Plain Text
4.2.5 Evaluation of Dialogue Effectiveness and Coverage
4.3 Coordinating Questions and Answers
4.3.1 Learning Coordination Between a Request or Question and a Response
4.3.2 Computing Similarity Between Communicative Actions in Questions and Answers
4.4 A Social Promotion Chatbot
4.4.1 Communicating with Friends on Behalf of a Human Host
4.4.2 The Domain of Social Promotion
4.4.3 The Chatbot Architecture
4.4.4 Use Cases of CASP
4.4.5 Evaluation
4.5 Improving Discourse Parsing
4.5.1 Syntactic Generalization of a Sentence Being Parsed and an AMR Template
4.5.2 Rhetorical Relation Enhancement Algorithm
4.5.3 Generalization Levels: From Syntax to Semantics to Discourse
4.5.4 Evaluation
4.6 Related Work and Conclusions
4.6.1 Constructing Dialogues from Plain Text
4.6.2 Conclusions on CASP
References
5 Concluding a CRM Session
5.1 Concluding a Question Answering Session
5.1.1 Building a Structure of Conclusive Answer
5.1.2 Content Compilation Algorithm
5.1.3 A Log of Answer Generation
5.1.4 Modeling the Content Structure of Texts
5.1.5 Building Answer Document Based on Similarity and Compositional Semantics
5.2 Defeating Conclusion of a Support Session
5.2.1 Introducing Defeating Reply
5.2.2 An Algorithm for Identifying Answers with Defeating Arguments
5.2.3 Representing Nested Arguments by R-C Framework
5.2.4 Reasoning with Arguments Extracted from Text
5.2.5 Adjusting Available Discourse Parsers to Argumentation Domain
5.2.6 Evaluation
5.3 Discussion and Conclusions
References
6 Truth, Lie and Hypocrisy
6.1 Anatomy of a Lie
6.1.1 Introduction: A Discourse of a Lie
6.1.2 Example of Misrepresentation in User-Generated Content
6.1.3 Example of Misrepresentation in Professional Writing
6.1.4 Background and Related Work
6.1.5 Dataset Description
6.1.6 Communicative Discourse Trees to Represent Truthfulness in Text
6.1.7 Evaluation
6.1.8 Two Dimensions of Lie Detection
6.1.9 Fact-Checking Tools
6.1.10 Conclusions
6.2 Detecting Hypocrisy in Company and Customer Communication
6.2.1 Introducing Hypocrisy
6.2.2 Hypocrisy in Customer Complaints
6.2.3 Building a Dataset of Sentences with Hypocrisy
6.2.4 Templates for Sentences with Hypocrisy
6.2.5 Assessing Coordination of Prominent Entities
6.2.6 Hypocrisy in Tweets
6.2.7 Expressing Hypocrisy in a Dialogue
6.2.8 System Architecture
6.2.9 Evaluation
6.2.10 Related Work and Discussions
6.2.11 Hypocrysy versus Controversy Stance, Sarcasm, Sentiments
6.2.12 Measuring Contention Between Say and Do Parts
6.2.13 Hypocrisy and Opinion Formation
6.2.14 Conclusions
6.3 Detecting Rumor and Disinformation by Web Mining
6.3.1 Introduction
6.3.2 Definitions and Examples
6.3.3 Examples of Disinformation as Entity Substitutions
6.3.4 Disinformation and Rumor Detection Algorithm
6.3.5 Evaluation
6.3.6 Related Work and Conclusions
6.3.7 Corruption Networks
6.3.8 Lying at Work
References
7 Reasoning for Resolving Customer Complaints
7.1 Introduction
7.1.1 Why Are Both the Deductive and Inductive Components Required?
7.1.2 Statistical or Deterministic Machine Learning?
7.2 The System Architecture
7.3 Inductive Machine Learning as a Logic Program
7.4 Merging Deductive and Inductive Reasoning About Action
7.5 Predicting Inter-Human Interactions in Customer Complaints
7.5.1 Introducing the Domain of Customers’ Complaints
7.5.2 Selecting the Features, Fluents and Actions
7.5.3 Setting the Learning Environment
7.5.4 Further Classification of Complaint Scenarios
7.5.5 Applying Semantic Templates
7.5.6 Evaluation of Prediction Results
7.6 Conclusions
References
8 Concept-Based Learning of Complainants’ Behavior
8.1 Introduction
8.2 Logical Simulation of the Behavior
8.3 Complaint Validity, Complaint Management and CRM
8.4 Complaint Scenario and Communicative Actions
8.5 Formalizing Conflict Scenarios
8.6 Semantics of Communicative Actions
8.7 Defining Scenarios as Graphs and Learning Them
8.8 Assigning a Scenario to a Class
8.9 JSM Learning in Terms of Formal Concept Analysis
8.10 Finding Similarity Between Scenarios
8.11 Scenarios as Sequences of Local Logics
8.12 Evaluation
8.12.1 Assessing Validity of Banking Complaints
8.13 Assessing Validity of Travelers’ Complaints
8.14 Using ComplaintEngine
8.15 Selecting Products by Features Using Customer Feedback
8.16 Discussion and Conclusions
References
9 Reasoning and Simulation of Mental Attitudes of a Customer
9.1 Introduction
9.1.1 The Task of the ToM Engine
9.2 A Model of a Mental Attitude of a Customer
9.2.1 Mental States and Actions
9.2.2 An Example of a Definition of a Mental Action
9.2.3 Derived Metapredicates
9.2.4 Handling Multiple Meanings
9.2.5 Representing Emotions
9.3 Simulating Reasoning About the Mental States
9.4 Implementation of Simulation
9.4.1 Choosing the Best Action Taking into Account Yourself Only
9.4.2 Choosing the Best Action Considering an Action Selection by Others
9.4.3 The Repository of Behaviors
9.5 Evaluation of the ToM Engine
9.5.1 Evaluation of Precision
9.5.2 Evaluation of Completeness
9.5.3 Evaluation of Complexity
9.6 Introduction to Meta-Reasoning and Introspection of ToM Engine
9.6.1 Meta-Interpreter of NL
9.6.2 Metaprogramming Tricks for Q/A
9.7 ToM Engine Support for Customer Complaint Processing
9.7.1 Linked Subscenarios
9.8 Front End of ToM Engine
9.8.1 Related Systems
9.9 Discussion and Conclusions
References
10 CRM Becomes Seriously Ill
10.1 Introduction
10.2 Defining DI
10.3 Companies Sick with Distributed Incompetence
10.3.1 Managing Distributed Incompetence Organizations
10.3.2 Whistleblowing in Distributed Incompetence Organizations
10.3.3 The Financial Crisis and Distributed Incompetence Organizations
10.3.4 Distributed Incompetence and Competitive Rating
10.3.5 Irrationality of Agents Under Distributed Incompetence
10.3.6 Aggressive DI
10.3.7 Machine Learning of DI
10.4 Detecting DI in Text
10.4.1 Distributed Incompetence and Rhetorical Relations
10.4.2 Semantic Cases of Distributed Incompetence
10.4.3 A Detection Dataset
10.4.4 Discourse-Level Features
10.4.5 Implementation of the Detector of Distributed Incompetence
10.4.6 Detection Results
10.5 Customer Service and Covid-19
10.6 Conclusions: Curing Distributed Incompetence
References
11 Conclusions