TOPSIS with Multiple Linear Regression for Multi-Document Text Summarization

Abstract

The huge amount of information in the internet makes rapid need of text
summarization. Text summarization is the process of selecting important sentences
from documents with keeping the main idea of the original documents. This paper
proposes a method depends on Technique for Order of Preference by Similarity to
Ideal Solution (TOPSIS). The first step in our model is based on extracting seven
features for each sentence in the documents set. Multiple Linear Regression (MLR)
is then used to assign a weight for the selected features. Then TOPSIS method
applied to rank the sentences. The sentences with high scores will be selected to be
included in the generated summary. The proposed model is evaluated using dataset
supplied by the Text Analysis Conference (TAC-2011) for English documents. The
performance of the proposed model is evaluated using Recall-Oriented Understudy
for Gisting Evaluation (ROUGE) metric. The obtained results support the
effectiveness of the proposed model.