Automatic Text Summarization

Explore automatic text summarization techniques in this comprehensive 58-minute conference talk. Delve into the process of shortening text documents using software to create accurate and fluent summaries. Learn about extractive and abstractive approaches, examining common algorithms and tools used in the field. Discover methods for evaluating automated summaries, including precision, recall, utility, and the pyramid method. Gain insights into challenges like lack of balance and cohesion in extractive summaries. Investigate various techniques such as positional method, Luhn's method, Edmundson's method, and FRUMP. Understand the application of classification, maximal marginal relevance, and sequence-to-sequence models in text summarization. Acquire knowledge to determine what makes a good summary and how to assess its quality effectively.

Intro
Automatic text summarization
Extractive vs. Abstractive Summary
Extractive Summaries-Lack of balance
Extractive Summaries-Lack of cohesion
Positional method
Luhn's method
Edmundson's method
FRUMP - Demonstration script
Classification
Maximal marginal relevance
Sequence to sequence
What makes a good summary?
Types of evaluation methods
Precision and Recall
Utility
Pyramid method