Research and Application of Automatic Text Summarization Technology Based on Deep Learning

It takes a lot of time and energy for users to obtain useful information from the massive data generated by the Internet. The text abstract is a refined expression of the content of the article, which can summarize the main content of the article. Text summarization technology can quickly allow users to obtain information that is valuable to them, and to a certain extent alleviate the problem of information overload in the era of big data. In this paper, we use the knowledge enhancement model to learn the semantic relationship of the real world by modeling the entity concept and other prior semantic knowledge in massive data, so as to overcome the disadvantage of using only the original language signal in the previous language model. Then the generative pre-training model is used to solve some specific problems in natural language generation, such as the exposure bias problem. The experimental results show that the model used in this paper works well on the Gigaword and CNN / DailyMail data sets. At the same time, the abstract generated on the nlpcc2017 Chinese abstract data has good accuracy and readability.