Paragraph-, word-, and coherence-based approaches to sentence ranking

Abstract

Sentence ranking is a crucial part of generating text summaries. We compared human sentence rankings obtained in a psycholinguistic experiment to three different approaches to sentence ranking: A simple paragraph-based approach intended as a baseline, two word-based approaches, and two coherence-based approaches. In the paragraph-based approach, sentences in the beginning of paragraphs received higher importance ratings than other sentences. The word-based approaches determined sentence rankings based on relative word frequencies (Luhn (1958); Salton & Buckley (1988)). Coherence-based approaches determined sentence rankings based on some property of the coherence structure of a text (Marcu (2000); Page et al. (1998)). Our results suggest poor performance for the simple paragraph-based approach, whereas wordbased approaches perform remarkably well. The best performance was achieved by a coherence-based approach where coherence structures are represented in a non-tree structure. Most approaches also outperformed the commercially available MSWord summarizer.

A Topic-driven Summarization using K-mean Clustering and Tf-Isf Sentence Ranking

“…Understanding of contents, reformulation and sentence-compression is done while abstraction [4,5] where as the sentences of text are ranked and important ones are picked-up in extraction. In extractive summarization, sentence/paragraph ranking [6,7,8] is the centre of attention; various methods are used to rank sentences/paragraphs. In [6], compare the effectiveness of paragraph, word, and coherence based sentence ranking approaches.…”

“…In extractive summarization, sentence/paragraph ranking [6,7,8] is the centre of attention; various methods are used to rank sentences/paragraphs. In [6], compare the effectiveness of paragraph, word, and coherence based sentence ranking approaches. The best performance was accomplished by coherence based approaches.…”

Discourse oriented summarization

“…The representation of discourse as a hierarchical tree is disputed by Wolf and Gibson (2005). They built a summarization system which used their non-hierarchical discourse model for content selection (Wolf and Gibson, 2004). They claim to have achieved a higher quality summaries with their graph-based approach than with the (RST) tree-based approach of Marcu (2000).…”

“…If each sentence in the source text is (manually) rated for importance, an extractive summary can be evaluated by calculating the sum of the ratings of its sentences Wolf and Gibson, 2004). An indirect method of rating sentences was used by (Otterbacher et al, 2005), who asked annotators to come up with a list of questions to information key to understanding the story.…”

“…For instance, Dang (2006) used the leading sentences of the most recent documents to create baseline summaries. Wolf and Gibson (2004) used the leading sentences of paragraphs in baseline summaries.…”