中文摘要：

设计一种基于潜在类别模型的新闻推荐模型，包含用户、新闻关键词和新闻类别三个外显变量及一个潜在类别变量。将用户、新闻和类别分别归类到相应的潜在类别中，根据用户兴趣偏好预测用户登陆新闻网站后可能访问的新闻项，生成个性化的新闻推荐序列，并通过实验证明了该三元模型的优越性。

英文摘要：

We present a news recommending model based on latent model, including three manifest variables, such as user, news items and news categories, and a latent variable, which classifies the manifest variables into corresponding latent categories. It can predict the news items users may read after gaining access to the news website and generate personalized news rankings according to users＆#39; prefer-ences. Experiments show the superiority of the triadic model.