Dynamic Multi-Document Summarization Research based on Matrix Subspace Analysis Model
Authors
Meiling Liu, Dequan Zheng, Tiejun Zhao, Honge Ren and Yang Yu
Abstract
In this paper, we described dynamic evolution of network information, as well as identify and
analysis the document collection on the same topic in different stages. Dynamic summarization considers the
different documents\u2019 temporal relationship in multi-document and analyzes the relationship between emerged
information and emerging information. In order to construct a dynamic evolution of content differences, a
dynamic multi-document summarization model was presented, called the Matrix Subspace Analysis Method
model. On this basis, proposed some efficient dynamic sentence weighting methods, and experiments on the
test data of Update Summarization in TAC2008, we showed effectiveness results.