n-grams of Seeds: A Hybrid System for Corpus-Based Text Summarization

Abstract

This paper presents a hybrid system for automatic text summarization which combines statistical and knowledge-based methods. In particular, it demonstrates how two corpus-based learning and indexing algorithms, namely an n-gram and a seed-oriented approach, may be combined to bring out the best of both approaches. This system selects sentences from an input text to constract a highly compressed, generic, and informative summary. The hybrid algorithm described here was developed and tested with a corpus of movie reviews collected from several on-line data bases.

Resources

Details