Research On The Application Of Image Content Recognition And Contrast Mining In Web Site Analysis

Today,as the quality of life improves,people travel more than ever before,and the demand for hotels increases.The boom in information technology has made social networking sites easy for hotel choices.However,because of the inability to conduct field visits to the hotels displayed on the website,it is a matter of wide spread to consumers about how to compare hotels to meet their needs.The images displayed on the website provide visual enjoyment while providing a good solution to this problem,because the images can reflect the actual situation of the hotel.But with more and more images on social networking sites,the problem of image information load is becoming more and more serious,which causes a lot of trouble for consumers to make artificial comparison choices.In view of the above problems,this paper after a lot of research,found that in the relevant research of social networking sites at home and abroad,most people tend to study for comments or ratings,but the contrast mining of image content is less research,so that image information can't be fully utilized.To achieve the difference between comparing two sets of large-scale image datasets and getting,this paper first uses deep learning techniques to identify entities in the image and get labels.Then,in order to reduce the adverse effect of information redundancy caused by entity similarity on the comparative analysis,the following methods are used: 1)the word vector conversion method in text analysis,vectorization of the entity label;2)the hierarchical clustering algorithm without specifying the number of categories Clustering of label vectors.Finally,the FP-Growth algorithm uses the FP-Growth algorithm to mine frequent item sets and compare the patterns of the vector set after clustering,and to identify the contrast ingesting mode in the two sets of image data sets.Analyze the differences between two sets of image datasets by identifying the contrast patterns.This article applies the above methods to famous social networking sites at home and abroad: TripAdvisor,Co-worker,Pinterest,etc.,and obtains relevant comparison models.These comparison models can provide a reliable reference for consumers and businesses to make consumption choices and formulate marketing strategies information.At the same time,the photo processing and contrast mining methods based on deep learning proposed in this article can also be applied to other social networks to help consumers and businesses provide effective consumption decisions and marketing strategies.