Practical Text Mining and Statistical Analysis for Non-structured Text Data Applications

Dedication

Endorsements for Practical Text Mining & Statistical Analysis for Non-structured Text Data Applications

Foreword 1

Foreword 2

Foreword 3

Acknowledgments

Preface

About the Authors

Introduction

Building the Workshop Manual

Communication

The Structure of this Book

Part I: Basic Text Mining Principles

Part II: Tutorials

Part III: Advanced Topics

Tutorials

Why Did We Write This Book?

What Are the Benefits of Text Mining?

Blast Off!

References

List of Tutorials by Guest Authors

Part I: Basic Text Mining Principles

Chapter 1. The History of Text Mining

Preamble

The Roots of Text Mining: Information Retrieval, Extraction, and Summarization

Information Extraction and Modern Text Mining

Major Innovations in Text Mining since 2000

The Development of Enabling Technology in Text Mining

Emerging Applications in Text Mining

Sentiment Analysis and Opinion Mining

IBM’s Watson: An “Intelligent” Text Mining Machine?

What’s Next?

Postscript

References

Chapter 2. The Seven Practice Areas of Text Analytics

Preamble

What is Text Mining?

The Seven Practice Areas of Text Analytics

Five Questions for Finding the Right Practice Area

The Seven Practice Areas in Depth

Interactions between the Practice Areas

Scope of This Book

Summary

Postscript

References

Chapter 3. Conceptual Foundations of Text Mining and Preprocessing Steps

Preamble

Introduction

Syntax versus Semantics

The Generalized Vector-Space Model

Preprocessing Text

Creating Vectors from Processed Text

Summary

Postscript

Reference

Chapter 4. Applications and Use Cases for Text Mining

Preamble

Why Is Text Mining Useful?

Extracting “Meaning” from Unstructured Text

Summarizing Text

Common Approaches to Extracting Meaning

Extracting Information through Statistical Natural Language Processing

Statistical Analysis of Dimensions of Meaning

Beyond Statistical Analysis of Word Frequencies: Parsing and Analyzing Syntax

Review

Improving Accuracy in Predictive Modeling

Using Statistical Natural Language Processing to Improve Lift

Using Dictionaries to Improve Prediction

Identifying Similarity and Relevance by Searching

Part of Speech Tagging and Entity Extraction

Summary

Postscript

References

Chapter 5. Text Mining Methodology

Preamble

Text Mining Applications

Cross-Industry Standard Process for Data Mining (CRISP-DM)

Example 1: An Exploratory Literature Survey Using Text Mining

Postscript

References

Chapter 6. Three Common Text Mining Software Tools

Preamble

Introduction

IBM SPSS Modeler Premium

SAS Text Miner

About the Scenarios in This SAS Section

Tips for Text Mining

STATISTICA Text Miner

Summary: STATISTICA Text Miner

Postscript

Part II: Introduction to the Tutorial and Case Study Section of This Book

Introduction

Reference

Tutorial AA. Case Study: Using the Social Share of Voice to Predict Events That Are about to Happen

Analysis

Summary

Tutorial BB. Mining Twitter for Airline Consumer Sentiment

Introduction

What Is R?

Loading Data into R

The twitteR Package

Extracting Text from Tweets

The plyr Package

Estimating Sentiment

Loading the Opinion Lexicon

Implementing Our Sentiment Scoring Algorithm

Algorithm Sanity Check

data.frames Hold Tabular Data

Scoring the Tweets

Repeat for Each Airline

Compare the Score Distributions

Ignore the Middle

Compare with ACSI’s Customer Satisfaction Index

Scrape the ACSI Website

Compare Twitter Results with ACSI Scores

Graph the Results

Notes and Acknowledgments

References

Tutorial A. Using STATISTICA Text Miner to Monitor and Predict Success of Marketing Campaigns Based on Social Media Data

Introduction

The Key Issue

Step 1: Collecting Data

Step 2: Monitoring the Situation

Step 3: Creating Predictive Models

Step 4: Performing a “What-If” Analysis of the Marketing Campaigns

Step 5: Performing Sentiment Analysis

Summary

Tutorial B. Text Mining Improves Model Performance in Predicting Airplane Flight Accident Outcome

Introduction

The Data

Text Mining the Data

Text Mining Results

Data Preparation

Using Text Mining Results to Build Predictive Models

Tutorial C. Insurance Industry: Text Analytics Adds “Lift” to Predictive Models with STATISTICA Text and Data Miner

Introduction

Data Description

Part A: Comparing the Lift of Predictive Models with and without Text Mining

Boosted Trees (without Text Material)

Boosted Trees Adding the Text Mining Variables

How to Merge Graphs

Part B: Enterprise Deployment

Summary

Tutorial D. Analysis of Survey Data for Establishing the “Best Medical Survey Instrument” Using Text Mining

Introduction

The Analysis

Summary

Tutorial E. Analysis of Survey Data for Establishing “Best Medical Survey Instrument” Using Text Mining: Central Asian (Russian Language) Study Tutorial 2: Potential for Constructing Instruments That Have Increased Validity

Introduction

The Analysis

Summary

Tutorial F. Using eBay Text for Predicting ATLAS Instrumental Learning

Introduction

Examining the Data by Types

Summary

Reference

Tutorial G. Text Mining for Patterns in Children’s Sleep Disorders Using STATISTICA Text Miner

Setting Up the Analysis

Reviewing Results

Summary

Tutorial H. Extracting Knowledge from Published Literature Using RapidMiner

Introduction

Motivation

A Brief Introduction to RapidMiner

Text Analytics in RapidMiner

Starting a New Process

Summary

Reference

Tutorial I. Text Mining Speech Samples: Can the Speech of Individuals Diagnosed with Schizophrenia Differentiate Them from Unaffected Controls?

Introduction

Objectives

Case Study: The Steps Used to Prepare the Data

Results and Analysis

Summary

References

Tutorial J. Text Mining Using STM™, CART®, and TreeNet® from Salford Systems: Analysis of 16,000 iPod Auctions on eBay

Installing the Salford Text Miner

Comments on the Challenge

Tutorial K. Predicting Micro Lending Loan Defaults Using SAS® Text Miner

Introduction

About SAS® Text Miner

Project Overview

Preparing the Data and Setting Up the Diagram

Creating a New Project

Registering the Table

Creating a New Diagram

Text Filter Node

Text Topic Node

Creating the Text Mining Flow

Inserting the Data

Understanding Text Parsing

Synonyms and Multiterm Words

Defining Topics

Other Uses of the Interactive Topic Viewer

Making the Predictive Model

Final Results

Viewing the Reports

Text Only Decision Tree

All Variable Text and Relational

Conclusion

Tutorial L. Opera Lyrics: Text Analytics Compared by the Composer and the Century of Composition—Wagner versus Puccini

Tutorial M. Case Study: Sentiment-Based Text Analytics to Better Predict Customer Satisfaction and Net Promoter® Score Using IBM®SPSS® Modeler

Introduction

Business Objectives

Case Study

Creating New Categories and Adding Missing Descriptors

Results and Analysis

Summary

References

Tutorial N. Case Study: Detecting Deception in Text with Freely Available Text and Data Mining Tools

Introduction

General Architecture for Test Engineering

Linguistic Inquiry and Word Count

Working with General Architecture for Test Engineering and Linguistic Inquiry and Word Count Output

Summary

References

Tutorial O. Predicting Box Office Success of Motion Pictures with Text Mining

Introduction

Analysis

Summary

References

Tutorial P. A Hands-On Tutorial of Text Mining in PASW: Clustering and Sentiment Analysis Using Tweets from Twitter

Introduction

Objective

Case Study

Categorization

Cluster Analysis

Analyzing Text Links

Additional Settings

Summary

Tutorial Q. A Hands-On Tutorial on Text Mining in SAS®: Analysis of Customer Comments for Clustering and Predictive Modeling

Introduction

Objective

Case Study

Summary

References

Tutorial R. Scoring Retention and Success of Incoming College Freshmen Using Text Analytics

Introduction

Part I. Predictive Modeling Using Only the Numeric Variables

Part II. Text Mining and Text Variables’ Word Frequencies and Concepts

Tutorial S. Searching for Relationships in Product Recall Data from the Consumer Product Safety Commission with STATISTICA Text Miner

Specifying the Analysis

Reviewing the Results

Tutorial T. Potential Problems That Can Arise in Text Mining: Example Using NALL Aviation Data

Introduction

Spelling Errors

Example: Finding Spelling Errors in Text Miner

Combine Words

Misspellings as Synonyms

Unexpected Terms

Example: Finding Unexpected Terms

Different File Types

Summary

Tutorial U. Exploring the Unabomber Manifesto Using Text Miner

Introduction

Summarizing the Text

Searching for Trends with Pronouns

References

Tutorial V. Text Mining PubMed: Extracting Publications on Genes and Genetic Markers Associated with Migraine Headaches from PubMed Abstracts

Tutorial W. Case Study: The Problem with the Use of Medical Abbreviations by Physicians and Health Care Providers

The Present Problem in the use of Medical Abbreviations by Physicians and Health Care Providers

TJC (JCAHO) “Do Not Use” Abbreviations

Additional Abbreviations, Acronyms, and Symbols

Using the “Text Mining Project” Format of STATISTICA Text Miner

Using TextMiner3.dbs

Conclusion

Intervention Training Needed

References

Tutorial X. Classifying Documents with Respect to “Earnings” and Then Making a Predictive Model for the Target Variable Using Decision Trees, MARSplines, Naïve Bayes Classifier, and K-Nearest Neighbors with STATISTICA Text Miner

Introduction: Automatic Text Classification

Data File with File References

Specifying the Analysis

Processing the Data Analysis

Saving the Extracted Word Frequencies to the Input File

Initial Feature Selection

General Classification and Regression Trees

K-Nearest Neighbors Modeling

Conclusion

Reference

Tutorial y. Case Study: Predicting Exposure of Social Messages: The Bin Laden Live Tweeter

Introduction

Analysis

Summary

Tutorial Z. The InFLUence Model: Web Crawling, Text Mining, and Predictive Analysis with 2010–2011 Influenza Guidelines—CDC, IDSA, WHO, and FMC

Abstract

Web Crawling and Text Mining of CDC Documents on FLU

Feature Selection

MARSplines Interactive Module Modeling

Boosted Trees

Naïve Bayes Modeling

K-Nearest Neighbors

Part III: Advanced Topics

Chapter 7. Text Classification and Categorization

Preamble

Introduction

Defining a Classification Problem

Feature Creation

Text Classification Algorithms

Combining Evidence

Evaluating Text Classifiers

Hierarchical Text Classification

Text Classification Applications

Summary

Postscript

References

Chapter 8. Prediction in Text Mining: The Data Mining Algorithms of Predictive Analytics

Preamble

Introduction

The Power of Simple Descriptive Statistics, Graphics, and Visual Text Mining

Visual Data Mining

Predictive Modeling (Supervised Learning)

Statistical Models versus General Predictive Modeling

Clustering (Unsupervised Learning)

Singular Value Decomposition, Principal Components Analysis, and Dimension Reduction

Association and Link Analysis

Summary

Postscript

References

Chapter 9. Entity Extraction

Preamble

Introduction

Text Features for Entity Extraction

Strategies for Entity Extraction

Choosing an Entity Extraction Approach

Evaluating Entity Extraction

Summary

Postscript

References

Chapter 10. Feature Selection and Dimensionality Reduction

Preamble

Introduction

Feature Selection

Feature Selection Approaches

Dimensionality Reduction

Linear Dimensionality Reduction Approaches

Postscript

References

Chapter 11. Singular Value Decomposition in Text Mining

Preamble

Introduction

Redundancy in Text

Dimensions of Meaning: Latent Semantic Indexing

The Math of Singular Value Decomposition

Graphical Representations and Simple Examples

Singular Value Decomposition in Equation Form

Singular Value Decomposition and Principal Components Analysis Eigenvalues

Some Practical Considerations

Extracting Dimensions

Subjective Methods: Reviewing Graphs

Analytical Methods: Building Models for Dimensions

Useful Analyses Based on Singular Value Decomposition Scores

Cluster Analysis

Predictive Modeling

When SVD Is Not Useful

Summary

Postscript

References

Chapter 12. Web Analytics and Web Mining

Preamble

Web Analytics

The Value of Web Analytics

The Future of Web Analytics and Web Mining

Postscript

References

Chapter 13. Clustering Words and Documents

Preamble

Introduction

Clustering Algorithms

Clustering Documents

Clustering Words

Cluster Visualization

Summary

Postscript

References

Chapter 14. Leveraging Text Mining in Property and Casualty Insurance

Preamble

Introduction

Property and Casualty Insurance as a Business

Analytics Opportunities in the Insurance Life Cycle

Driving Business Value Using Text Mining

Summary

Postscript

References

Chapter 15. Focused Web Crawling

Preamble

Introduction

The Focused Crawling Process

The Opportunities and Challenges of Mining the Web

Topic Hierarchies for Focused Crawling

Training the Document Classifier

Capturing User Feedback

Summary

Postscript

References

Chapter 16. The Future of Text and Web Analytics

Text Analytics and Text Mining

The Pros and Cons of Commercial Software versus Open Source Software

The Future of Text Mining

The Future of Web Analytics

Multisession Pathing

Integration of Web Analytics with Standard BI Tools

Attribution across Multiple Sessions

The Future: What Does It Hold?

New Areas That May Use Text Analytics in the Future

IBM Watson

Summary

References

IBM-Watson References

Chapter 17. Summary

Why Are You Reading This Chapter?

Our Perspective for Applying Text Mining Technology

Part I: Background and Theory

What Is Text Mining?

What Tools Can I Use?

Part II: The Text Mining Laboratory—28 Tutorials

Part III: Advanced Topics

Outlines of Chapter 7–15

Glossary

Index

How to Use the Data Sets and the Text Mining Software on the DVD or on Links for Practical Text Mining

I Data Sets for the Tutorials in Practical Text Mining

II SAS Text Miner Software

III Salford Systems Software, Including a New Text Miner Module Made for this Book (30-Day Free Trial Available)

IV STATISTICA Text Miner Software (30-day free trial on the DVD that accompanies this book)