English Document Query Automatic Abstracting Research

With the development of information technology and internet, people are located in the immense ocean of information. The information we can access turns to be more and more complex. People desire to index, abstract and condense the information and express main idea with fewer characters, in order to reduce the time of accessing information. Automatic text summarization offers such a solution, using computer to automatically abstract information that users need to save much time for users.Automatic text summarization has always been hot in the field of natural language processing. These years it tends to fuse with automatic question answering system while query-based summarization and related international evaluation conferences get more attention. Researches on semantic relations, discourse analysis and machine learning methods turn to be more and more. Under this background, this paper tries to make a detailed discussion to the important feature semantic relational triple and the application of machine learning methods in summarization filed.Semantic relation is one important feature of summarization methods. This paper proposes a new query-focused summarization method based on semantic relational triples. After syntax parsing, we abstract the semantic relational triples both from sentences and questions, then calculate the similarity of triples using search engine. The similarities of sentences and questions are then calculated to form the final question-focused summarization. Experiments on authoritative corpus show that this feature outperforms classical features and its results are comparable to international results.Extract summarization can be looked upon as a task of classification. This paper focuses on the classification method in automatic text summarization and applies it into automatic text summarization. The method transforms the process of automatic text summarization supported by training corpus into problem of two-class classification, and applies the mature supervised classification technologies to automatic text summarization. Experimentations show that the method effective. We also compared two classifiers and important characters like similarity to centroid and similarity to question.