Automatic Text Summarization for Dialogue Style

Automatic text summarization for dialogue style is a relatively new research area. Some key techniques are proposed in this paper, such as identifying whether or not the style of document is dialogue, identifying all of questions and their corresponding answers and linking them to form question-answer pairs. Due to the interactive nature of dialogue documents, local regions of coherence often stretch across different speakers. An approach to automatically identify those regions of local coherence is presented. Firstly, we identify all question paragraphs in a given document. Secondly, we detect the answer paragraphs of each question paragraph and exploit text segmentation to form question-answer pairs. Lastly, we extract some sentences from the generic content and the question-answer pairs to generate a complete summary. Experimental results show that our approach is high efficient and improves significantly the coherency of the summary while not compromising informativeness