Multi-Document Extractive Summarization Using Window-Based Sentence Representation

Multi-document summarization has gained popularity in many real world applications because significant information can be obtained within a short time. Extractive summarization aims to generate a summary of a document or a set of documents by ranking sentences, whose performance relies heavily on the quality of sentence features. However, almost all previous algorithms require hand-crafted features for sentence representation. In this paper, we leverage on word embedding to represent sentences so as to avoid the intensive labor of feature engineering. We propose a new technique, namely window-based sentence representation (WSR), to obtain the features of sentences using pre-trained word vectors. The method is developed based on the Extreme Learning Machine (ELM). Our proposed framework does not require any prior knowledge and therefore can be applied to various document summarization tasks with different languages, written styles and so on. We evaluate our proposed method on the DUC 2006 and 2007 datasets. This proposed method achieves superior performance compared with state-of-the-arts document summarization algorithms with a much faster learning speed.