The book is concerned with contemporary methodologies used for automatic text summarization. It proposes interesting approaches to solve well-known problems on text summarization using Computational Intelligence (CI) techniques including cognitive approaches. This book offers a thorough examination of the state-of-the-art methods to describe text summarization. For both extractive summarizing tasks and abstractive summary tasks, the reader will discover in-depth treatment of several methodologies utilizing Machine Learning (ML), Natural Language Processing (NLP), and data mining techniques. Additionally, it is shown how summarizing methodologies can be used in a variety of applications, including healthcare and social media domain along with the possible research directions and future scope. NLTK is an acronym for Natural Language Toolkit. It is the most commonly used Python package for handling human language data. It includes libraries for categorization, tokenization, stemming, tagging, parsing, and other text processing tasks. For text summarization, the NLTK employs the TF-IDF approach.
Author(s): V. Priya, K. Umamaheswari
Publisher: CRC Press
Year: 2023
Language: english
Pages: 229
Cover
Half Title
Title Page
Copyright Page
Table of Contents
About This Book
Preface
Chapter 1 Concepts of Text Summarization
1.1 Introduction
1.2 Need for Text Summarization
1.3 Approaches to Text Summarization
1.3.1 Extractive Summarization
1.3.2 Abstractive Summarization
1.4 Text Modeling for Extractive Summarization
1.4.1 Bag-of-Words Model
1.4.2 Vector Space Model
1.4.3 Topic Representation Schemes
1.4.4 Real-Valued Model
1.5 Preprocessing for Extractive Summarization
1.6 Emerging Techniques for Summarization
1.7 Scope of the Book
References
Sample Code
Sample Screenshots
Chapter 2 Large-Scale Summarization Using Machine Learning Approach
2.1 Scaling to Summarize Large Text
2.2 Machine Learning Approaches
2.2.1 Different Approaches for Modeling Text Summarization Problem
2.2.2 Classification as Text Summarization
2.2.2.1 Data Representation
2.2.2.2 Text Feature Extraction
2.2.2.3 Classification Techniques
2.2.3 Clustering as Text Summarization
2.2.4 Deep Learning Approach for Text Summarization
References
Sample Code
Chapter 3 Sentiment Analysis Approach to Text Summarization
3.1 Introduction
3.2 Sentiment Analysis: Overview
3.2.1 Sentiment Extraction and Summarization
3.2.1.1 Sentiment Extraction from Text
3.2.1.2 Classification
3.2.1.3 Score Computation
3.2.1.4 Summary Generation
3.2.2 Sentiment Summarization: An Illustration
Summarized Output
3.2.3 Methodologies for Sentiment Summarization
3.3 Implications of Sentiments in Text Summarization
Cognition-Based Sentiment Analysis and Summarization
3.4 Summary
Practical Examples
Example 1
Example 2
Sample Code (Run Using GraphLab)
Example 3
References
Sample Code
Chapter 4 Text Summarization Using Parallel Processing Approach
4.1 Introduction
Parallelizing Computational Tasks
Parallelizing for Distributed Data
4.2 Parallel Processing Approaches
4.2.1 Parallel Algorithms for Text Summarization
4.2.2 Parallel Bisection k-Means Method
4.3 Parallel Data Processing Algorithms for Large-Scale Summarization
4.3.1 Designing MapReduce Algorithm for Text Summarization
4.3.2 Key Concepts in Mapper
4.3.3 Key Concepts in Reducer
4.3.4 Summary Generation
An Illustrative Example for MapReduce
Good Time: Movie Review
4.4 Other MR-Based Methods
4.5 Summary
4.6 Examples
K-Means Clustering Using MapReduce
Parallel LDA Example (Using Gensim Package)
Sample Code: (Using Gensim Package)
Example: Creating an Inverted Index
Example: Relational Algebra (Table JOIN)
References
Sample Code
Chapter 5 Optimization Approaches for Text Summarization
5.1 Introduction
5.2 Optimization for Summarization
5.2.1 Modeling Text Summarization as Optimization Problem
5.2.2 Various Approaches for Optimization
5.3 Formulation of Various Approaches
5.3.1 Sentence Ranking Approach
5.3.1.1 Stages and Illustration
5.3.2 Evolutionary Approaches
5.3.2.1 Stages
5.3.2.2 Demonstration
5.3.3 MapReduce-Based Approach
5.3.3.1 In-Node Optimization Illustration
5.3.4 Multi-objective-Based Approach
Summary
Exercises
References
Sample Code
Chapter 6 Performance Evaluation of Large-Scale Summarization Systems
6.1 Evaluation of Summaries
6.1.1 CNN Dataset
6.1.2 Daily Mail Dataset
6.1.3 Description
6.2 Methodologies
6.2.1 Intrinsic Methods
6.2.2 Extrinsic Methods
6.3 Intrinsic Methods
6.3.1 Text Quality Measures
6.3.1.1 Grammaticality
6.3.1.2 Non-redundancy
6.3.1.3 Reverential Clarity
6.3.1.4 Structure and Coherence
6.3.2 Co-selection-Based Methods
6.3.2.1 Precision, Recall, and F-score
6.3.2.2 Relative Utility
6.3.3 Content-Based Methods
6.3.3.1 Content-Based Measures
6.3.3.2 Cosine Similarity
6.3.3.3 Unit Overlap
6.3.3.4 Longest Common Subsequence
6.3.3.5 N-Gram Co-occurrence Statistics: ROUGE
6.3.3.6 Pyramids
6.3.3.7 LSA-Based Measure
6.3.3.8 Main Topic Similarity
6.3.3.9 Term Significance Similarity
6.4 Extrinsic Methods
6.4.1 Document Categorization
6.4.1.1 Information Retrieval
6.4.1.2 Question Answering
6.4.2 Summary
6.4.3 Examples
Bibliography
Chapter 7 Applications and Future Directions
7.1 Possible Directions in Modeling Text Summarization
7.2 Scope of Summarization Systems in Different Applications
7.3 Healthcare Domain
Future Directions for Medical Document Summarization
7.4 Social Media
Challenges in Social Media Text Summarization
Domain Knowledge and Transfer Learning
Online Learning
Information Credibility
Applications of Deep Learning
Implicit and Explicit Information for Actionable Insights
7.5 Research Directions for Text Summarization
7.6 Further Scope of Research on Large-Scale Summarization
Conclusion
References
Appendix A: Python Projects and Useful Links on Text Summarization
Appendix B: Solutions to Selected Exercises
Index
