Research And Application Of Text Recommendation Method Based On Automatic Summarization Technology

Internet text messages tend to be massive and unstructured,quickly grasp the main thrust of the article from the mass text is the urgent needs of users.Automatic summary technology extracts the main content of the article so that users can understand the main topics of the article in a short time and obtain the valuable information.By using the text recommendation with automatic summary technology can quickly focus on the topic of the article and improve the recommendation efficiency for the target user.The main contents of this paper include the following aspects:(1)Automatically generated text summaries using graph model methods to obtain the main components of the articles,and reduce dimensions of the recommended algorithm.In the construction of the graph model,the similarity of the sentences is calculated based on word vectors and there are three ways to calculate the similarity of sentences,which are word vectors based on the average feature,the maximum similarity of feature words and the average similarity of feature words.Compared with the traditional methods that based on information quantity,the experimental results show that our approach is better than the former.(2)Optimized the convolutional neural network model and realized the text classification.It can accelerate the training time and improve the classification accuracy by training the result of the automatic summarization.(3)Acquire users' behavior attributes to establish user interest model.By analyzing the behavior of users clicking articles,reading speed,downloading articles and so on,the user's interest in the text is extracted.Extracted the users interest in the text by analyzing the behaviors of users that click articles,the speed of reading articles and download the article.According to the classification information of the text,the user's degree of interest in a single piece of text is mapped to the preference of the category which the text belongs to and established the user interest model.Therefore,according to the matchingdegree of user interest model and text classification features,we can recommend articles to users.(4)Designed and implemented a text recommendation system based on automatic summarization technology.The system focused on three categories information that are weapons and equipment,the Chinese military intelligence and the international military intelligence.The original information automatically generated summary which were sent into the convolution neural network model classification,and the results are automatically written to the corresponding form in database.According to the different operation behaviors of the users,the system will eventually generate corresponding user interest models and implement personalized text recommendation...