A dynamic programming model for text segmentation based on min-max similarity

Text segmentation has a wide range of applications such as information retrieval, question answering and text summarization. In recent years, the use of semantics has been proven to be effective in improving the performance of text segmentation. Particularly, in finding the subtopic boundaries, there have been efforts in focusing on either maximizing the lexical similarity within a segment or minimizing the similarity between adjacent segments. However, no optimal solutions have been attempted to simultaneously achieve maximum within-segment similarity and minimum between-segment similarity. In this paper, a domain independent model based on min-max similarity (MMS) is proposed in order to fill the void. Dynamic programming is adopted to achieve global optimization of the segmentation criterion function. Comparative experimental results on real corpus have shown that MMS model outperforms previous segmentation approaches .