Context-Aware Machine Learning and Mobile Data Analytics: Automated Rule-based Services with Intelligent Decision-Making

Part I Preliminaries

1 Introduction to Context-Aware Machine Learning and Mobile Data

Analytics

1.1 Introduction

1.2 Context-Aware Machine Learning

1.3 Mobile Data Analytics

1.4 An Overview of this Book

1.5 Conclusion

References

2 Application Scenarios and Basic Structure for Context-Aware

Machine Learning Framework

2.1 Motivational Examples with Application Scenarios

2.2 Structure and Elements of Context-Aware Machine Learning

Framework

2.2.1 Contextual Data Acquisition

2.2.2 Context Discretization

2.2.3 Contextual Rule Discovery

2.2.4 Dynamic Updating and Management of Rules

2.3 Conclusion

References

3 A Literature Review on Context-Aware Machine Learning and

Mobile Data Analytics

3.1 Contextual Information

3.1.1 Definitions of Contexts

3.1.2 Understanding the Relevancy of Contexts

3.2 Context Discretization

3.2.1 Discretization of Time-Series Data

3.2.2 Static Segmentation

vii

viii Contents

3.2.3 Dynamic Segmentation

3.3 Rule Discovery

3.3.1 Association Rule Mining

3.3.2 Classification Rules

3.4 Incremental Learning and Updating

3.5 Identifying the Scope of Research

3.6 Conclusion

References

Part II Context-Aware Rule Learning and Management

4 Contextual Mobile Datasets, Pre-processing and Feature Selection

4.1 Smart Mobile Phone Data and Associated Contexts

4.1.1 Phone Call Log

4.1.2 Mobile SMS Log

4.1.3 Smartphone App Usage Log

4.1.4 Mobile Phone Notification Log

4.1.5 Web or Navigation Log

4.1.6 Game Log

4.1.7 Smartphone Life Log

4.1.8 Dataset Summary

4.2 Examples of Contextual Mobile Phone Data

4.2.1 Time-Series Mobile Phone Data

4.2.2 Mobile phone data with multi-dimensional contexts

4.2.3 Contextual Apps Usage Data

4.3 Data Preprocessing

4.3.1 Data Cleaning

4.3.2 Data Integration

4.3.3 Data Transformation

4.3.4 Data Reduction

4.4 Dimensionality Reduction

4.4.1 Feature Selection

4.4.2 Feature Extraction

4.4.3 Dimensionality Reduction Algorithms

4.5 Conclusion

References

5 Discretization of Time-Series Behavioral Data and Rule Generation

based on Temporal Context

5.1 Introduction

5.2 Requirements Analysis

5.3 Time-series Segmentation Approach

5.3.1 Approach Overview

5.3.2 Initial Time Slices Generation

5.3.3 Behavior-Oriented Segments Generation

Contents ix

5.3.4 Selection of Optimal Segmentation

5.3.5 Temporal Behavior Rule Generation using Time Segments

5.4 Effectiveness Comparison

5.5 Conclusion

References

6 Discovering User Behavioral Rules based on Multi-dimensional

Contexts

6.1 Introduction

6.2 Multi-dimensional Contexts in User Behavioral Rules

6.3 Requirements Analysis

6.4 Rule Mining Methodology

6.4.1 Identifying the Precedence of Context

6.4.2 Designing Association Generation Tree

6.4.3 Extracting Non-Redundant Behavioral Association Rules

6.5 Experimental Analysis

6.5.1 Effect on the Number of Produced Rules

6.5.2 Effect of Confidence Preference the Predicted Accuracy

6.5.3 Effectiveness Comparison

6.6 Conclusion

References

7 Recency-based Updating and Dynamic Management of Contextual

Rules

7.1 Introduction

7.2 Requirements Analysis

7.3 An Example of Recent Data

7.4 Identifying Optimal Period of Recent Log Data

7.4.1 Data Splitting