Research And Implementation Of Chinese Single Document Automatic Summarization

Since the Automatic Digest task has been put forward for nearly 70 years,the researchers have maintained a wide range of ongoing attention,especially after entering the internet age of information explosion,the research heat based on automatic abstraction is constantly improving,the related results have been published,and the automatic ion technology has been greatly developed.There are two main types of automatic s:extraction and generation.The automatic abstraction of extraction type is composed of important fragments of the original text,such as sentences,paragraphs and so on.The generated automatic Digest is a direct exposition of its main thrust according to the original text.At present,the mainstream is the extraction summary,which has a good adaptability to multilingual,multi-domain,multi-style document sets.Generative methods often use more complex natural language comprehension and natural language generation and other technologies,from the current research results are not ideal,the practicality is poor.The purpose of this paper is to study and design the extraction automatic abstraction system based on machine learning for a single document,and to complete the automatic ion task.Based on the introduction of Word vector features of deep learning and the characteristics of various traditional sentence level features and Word levels,the feature combinations of these machine learning algorithms are carried out by using different collaborative algorithms,and the experimental results show that the performance of different collaborative algorithms varies after feature combination,but the performance of automatic abstracts is effectively improved.