Research on Automatic Summarization System Based on Topic Partition

Automatic summarization is an important issue in natural language processing and becomes more and more important in some domains such as web information retrieval. This paper describes an automatic summarization system based on topic partition: Firstly, it analyzes the text structure of the web pages, then proposes an algorithm for multi-topic text partitioning based on semantic distance between paragraphs, and finally get the summarization by extracting the key sentences from different topics. This method can makes the abstract of multi-topic article have more general content and more balanced structure. The experimental result shows that its value is stable.