Multi-Document Summarization

Dingding Wang

Introduction

Multi-document summarization is the process of generating a short summary by reducing documents in size while retaining the main characteristics of the original documents.

Types of Summaries

• Form - Abstraction v.s. Extraction
• Context - Generic summaries v.s. Query relevant summaries
• Purpose - Traditional summaries, Comparative summaries, Evolutionary summaries, Update summaries, Review summaries, etc.

Related Papers

• Centroid-based methods - MEAD, NeATS, XDoX
• Graph-based methods - LexPageRank, Mihalcea05, Wan08_1Wan08_2
• Clustering and factorization based methods - Park08, Wang08_1, Wang08_2, Wang09_1, Wang09_2
• Other methods -LSA, CRF, MMR, HMM, DUC publications

Summarization Demos

• Copernic
• MEAD
• LexRank
• SumView
• Dingding's Summarizer

Resources

• Document Understanding Conference (DUC)/Text Analysis Conference (TAC)
• A tutorial in SIGIR 2003 and 2004
• summarization.com
• Google scholar search results (09/09/2009)