In this paper, we present a method for extractive multi-document summarization using a hybrid approach for sentence scoring that combines the benefits of regression model and topic model. Fusing the regression model based score with topic model based score combines the benefit of both methods for ranking sentences and words that are scored on the basis of both surface and topical features. We use support vector regression based model for obtaining sentence scores and sparseTLM topic model for obtaining word scores. Both sentence and word scores are combined using BiRank algorithm for sentence ranking. An Integer Linear Programming method is used to select summary sentences maximizing coverage of summary based on ROUGE scores. The proposed method is shown to outperform existing state-of-the-art methods.

Text Summarization, Topic Model, Support Vector Regression, BiRank Algorithm