论文部分内容阅读
【目的】社交网络环境下的用户兴趣建模是好友推荐、精准营销的关键,利用微博用户分享的图像,提出一种基于图像语义的用户兴趣建模方法,旨在更加准确地预测用户的真实兴趣。【方法】在获取新浪微博用户图像数据的基础上,使用图像的高层语义表达用户兴趣特征,基于这些特征使用SVM训练得到图像语义分类器进行预测。【结果】实验结果表明,本文建立的模型能够较为准确地预测用户真实兴趣,169位用户分类的准确率达到97.38%,召回率为98.92%,F值为98.14%。【局限】由于实验图像数据集有限,未能完整地覆盖用户所有的兴趣类别。【结论】该模型能够基于用户分享的图像较为准确地预测用户兴趣,表明了图像高层语义的有效性,同时为图像高层语义应用研究提供了一定的理论和技术基础。
【Objective】 The modeling of user interest in social network environment is the key to friend recommendation and precision marketing. Based on the images shared by Weibo users, a user interest modeling method based on image semantics is proposed to predict the user more accurately Real interest. 【Method】 On the basis of obtaining the image data of Sina Weibo users, the user semantic features of high-level semantic representation of images are used to express the user interest features. Based on these features, the image semantic classifiers are predicted by SVM training. 【Result】 The experimental results show that the model established in this paper can predict the real interest of users more accurately. The accuracy of 169 users’ classification is 97.38%, the recall rate is 98.92% and the F value is 98.14%. [Limitations] Due to the limited experimental image dataset, all of the user’s interest categories were not completely covered. 【Conclusion】 The model can predict the user interest more accurately based on the images shared by users, indicating the validity of high-level semantics of images and providing theoretical and technical basis for the application of semantic high-level images.