Tarnowska / Ras / Daniel Recommender System for Improving Customer Loyalty

1. Auflage 2019
ISBN: 978-3-030-13438-9
Verlag: Springer International Publishing
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, Band 55, 133 Seiten

Reihe: Studies in Big Data

This book presents the Recommender System for Improving Customer Loyalty. New and innovative products have begun appearing from a wide variety of countries, which has increased the need to improve the customer experience. When a customer spends hundreds of thousands of dollars on a piece of equipment, keeping it running efficiently is critical to achieving the desired return on investment. Moreover, managers have discovered that delivering a better customer experience pays off in a number of ways. A study of publicly traded companies conducted by Watermark Consulting found that from 2007 to 2013, companies with a better customer service generated a total return to shareholders that was 26 points higher than the S&P 500. This is only one of many studies that illustrate the measurable value of providing a better service experience. The Recommender System presented here addresses several important issues. (1) It provides a decision framework to help managers determine which actions are likely to have the greatest impact on the Net Promoter Score. (2) The results are based on multiple clients. The data mining techniques employed in the Recommender System allow users to 'learn' from the experiences of others, without sharing proprietary information. This dramatically enhances the power of the system. (3) It supplements traditional text mining options. Text mining can be used to identify the frequency with which topics are mentioned, and the sentiment associated with a given topic. The Recommender System allows users to view specific, anonymous comments associated with actual customers. Studying these comments can provide highly accurate insights into the steps that can be taken to improve the customer experience. (4) Lastly, the system provides a sensitivity analysis feature. In some cases, certain actions can be more easily implemented than others. The Recommender System allows managers to 'weigh' these actions and determine which ones would have a greater impact.

Autoren/Hrsg.

Tarnowska, Katarzyna

Ras, Zbigniew W.

Daniel, Lynn

Inhaltsverzeichnis

1;Preface;6
2;About the Book;7
3;Contents;8
4;List of Figures;12
5;List of Tables;15
6;1 Introduction;17
6.1;1.1 Why Customer Experience Matters More Now?;17
6.2;1.2 Top (and Bottom) Line Reasons for Better Customer Experience;19
6.3;1.3 What is Next?;21
6.4;1.4 Final Observations;21
7;2 Customer Loyalty Improvement;23
7.1;2.1 Introduction to the Problem Area;23
7.2;2.2 Dataset Description;24
7.3;2.3 Decision Problem;25
7.4;2.4 Problem Area;25
7.4.1;2.4.1 Attribute Analysis;25
7.4.2;2.4.2 Attribute Reduction;26
7.4.3;2.4.3 Customer Satisfaction Analysis and Recognition;26
7.4.4;2.4.4 Providing Recommendations;27
7.5;Reference;27
8;3 State of the Art;28
8.1;3.1 Customer Satisfaction Software Tools;28
8.2;3.2 Customer Relationship Management Systems;29
8.3;3.3 Decision Support Systems;29
8.4;3.4 Recommender Systems;29
8.4.1;3.4.1 Recommender Systems for B2B;30
8.4.2;3.4.2 Types of Recommender Systems;31
8.4.3;3.4.3 Knowledge Based Approach for Recommendation;32
8.5;3.5 Text Analytics and Sentiment Analysis Tools;32
8.6;References;33
9;4 Background;35
9.1;4.1 Knowledge Discovery;35
9.1.1;4.1.1 Decision Reducts;35
9.1.2;4.1.2 Classification;37
9.1.3;4.1.3 Action Rules;38
9.1.4;4.1.4 Clustering;40
9.2;4.2 Text Mining;40
9.2.1;4.2.1 Sentiment Analysis;40
9.2.2;4.2.2 Aspect-Based Sentiment Analysis;41
9.2.3;4.2.3 Aspect Extraction;43
9.2.4;4.2.4 Polarity Calculation;45
9.2.5;4.2.5 Natural Language Processing Issues;46
9.2.6;4.2.6 Summary Generation;46
9.2.7;4.2.7 Visualizations;47
9.2.8;4.2.8 Measuring the Economic Impact of Sentiment;49
9.3;References;51
10;5 Overview of Recommender System Engine;54
10.1;5.1 High-Level Architecture;54
10.2;5.2 Data Preparation;56
10.2.1;5.2.1 Raw Data Import;56
10.2.2;5.2.2 Data Preprocessing;58
10.3;5.3 Semantic Similarity;62
10.4;5.4 Hierarchical Agglomerative Method for Improving NPS;64
10.5;5.5 Action Rules;66
10.6;5.6 Meta Actions and Triggering Mechanism;67
10.7;5.7 Text Mining;68
10.8;References;70
11;6 Visual Data Analysis;71
11.1;6.1 Decision Reducts as Heatmap;71
11.2;6.2 Classification Visualizations: Dual Scale Bar Chart and Confusion Matrix;73
11.3;6.3 Multiple Views;74
11.4;6.4 Evaluation Results;74
11.4.1;6.4.1 Single Client Data (Local) Analysis;75
11.4.2;6.4.2 Global Customer Sentiment Analysis and Prediction;76
11.5;6.5 User-Friendly Interface for the Recommender System;77
12;7 Improving Performance of Knowledge Miner;80
12.1;7.1 Introduction;80
12.2;7.2 Problem Statement;80
12.3;7.3 Assumptions;81
12.4;7.4 Strategy and Overall Approach;82
12.5;7.5 Evaluation;84
12.5.1;7.5.1 Experimental Setup;84
12.5.2;7.5.2 Results;85
12.5.3;7.5.3 New Rule Format in RS;89
12.6;7.6 Plans for Remaining Challenges;96
12.7;Reference;96
13;8 Recommender System Based on Unstructured Data;97
13.1;8.1 Introduction;97
13.2;8.2 Problem Statement;97
13.3;8.3 Assumptions;97
13.4;8.4 Strategy and Overall Approach;99
13.4.1;8.4.1 Data Transformation;99
13.4.2;8.4.2 Action Rule Mining;100
13.4.3;8.4.3 Ideas for the Improvement of Opinion Mining;101
13.4.4;8.4.4 Sentiment Extraction;101
13.4.5;8.4.5 Polarity Calculation;102
13.5;8.5 Evaluation;103
13.5.1;8.5.1 Initial Experiments;103
13.5.2;8.5.2 Experimental Setup;103
13.5.3;8.5.3 Improving Sentiment Analysis Algorithm;104
13.5.4;8.5.4 Experimental Results;108
13.5.5;8.5.5 Modified Algorithm for Opinion Mining;110
13.5.6;8.5.6 Comparing Recommendations with the Previous Approach;112
13.6;8.6 Plans for Remaining Challenges;116
13.6.1;8.6.1 Complex and Comparative Sentences;117
13.6.2;8.6.2 Implicit Opinions;118
13.6.3;8.6.3 Feature and Opinion in One Word;118
13.6.4;8.6.4 Opinion Words in Different Context;119
13.6.5;8.6.5 Ambiguity;119
13.6.6;8.6.6 Misspellings;120
13.6.7;8.6.7 Phrases, Idiomatic and Phrasal Verbs Expressions;120
13.6.8;8.6.8 Entity Recognition From Pronouns and Names;120
13.7;References;121
14;9 Customer Attrition Problem;122
14.1;9.1 Introduction;122
14.2;9.2 Problem Statement;122
14.3;9.3 Assumptions;124
14.4;9.4 Strategy and Overall Approach;124
14.4.1;9.4.1 Automatic Data Labelling;124
14.4.2;9.4.2 Pattern Mining;125
14.4.3;9.4.3 Sequence Mining;126
14.4.4;9.4.4 Action Rule, Meta Action Mining and Triggering;126
14.5;9.5 Evaluation;126
14.5.1;9.5.1 Initial Data Analysis;127
14.5.2;9.5.2 Attribute Selection;127
14.5.3;9.5.3 Classification Model;128
14.5.4;9.5.4 Action Rule Mining;129
14.6;9.6 Plans for Remaining Challenges;131
14.7;Reference;131
15;10 Conclusions;132
15.1;10.1 Contribution;132
15.2;10.2 Future Work;133