| With the rapid development of Internet technology in China,the scale of mobile news users is increasing year by year,and the user's demand experience is constantly improving,which puts forward higher requirements for the use of digital technology to produce and process news.How to produce large quantities of accurate news from explosive and disorderly news sources at low cost requires changes in the production mode of Internet news information.Based on the machine learning algorithm and the method of news information production,this thesis proposes a method of automatic recommendation pictures matching for news based on the fusion of multiple algorithms.Firstly,the logical regression classifier based on the word bag model is used to classify the news text into the news text with words on the image and the news text without words on the image.Then,according to different prediction results,the cross-fusion of two-type text similarity calculation methods and the BP(Back Propagation)neural network based matching recommendation are carried out respectively.The former calculates the similarity of Manhattan distance based on literal distance similarity and LSI(Latent Semantic Indexing)algorithm based on semantic similarity,and fuses the results to automatically recommend 10 pictures that match the text content.In thelatter,image and text features are mosaic as input features to predict the relationship between image and text.Finally,according to the prediction relationship,users automatically recommend the 10 most matching pictures with text content.This method can reduce the time of manual matching and proofreading in the process of news processing and production,promote the development of "machine + manual" production mode of news information,and further promote the study of the relationship between pictures and texts. |
PDF Full Text Request