Automatic multi-document summarization using a variable-based framework.

• Main Author: Ou, Shiyan.
• Other Authors: Khoo, Christopher Soo Guan
• Format: Theses and Dissertations
• Published: 2008
• Subjects: DRNTU::Library and information science::Libraries::Automation
• Online Access: http://hdl.handle.net/10356/1660
• Institution: Nanyang Technological University

This study was in the area of automatic multi-document summarization. The objective of this study was to develop a method for the automatic construction of summaries of sociology dissertation abstracts, focusing on research concepts and their research relationships. A variable-based framework was proposed to integrate and organize the four kinds of information extracted from different dissertation abstracts, i.e. research concepts and their research relationships, contextual relations and research methods. A hybrid summarization method involving both extraction and abstraction techniques was developed. It focused more on extraction and integration of research concepts and their research relationships across dissertation abstracts and presenting the integrated information in a Web-based interface. The evaluation results indicated that the variable-based summaries were preferred by the majority of researchers and general users, to the summaries generated by the sentence-extraction methods.