摘要
情境信息是一个影响人的兴趣的重要因素。在传统LDA模型的基础上加入情境信息对LDA模型的结果进行调整。在用LDA模型生成的文档-主题和主题-词的基础上,将用户兴趣根据不同的情境信息进行划分,进一步生成主题-心情分布。在此基础上提出基于时间情境的Time-LDA算法和基于心情情境的Mood-LDA算法。在真实的数据集上的实验表明所提出的算法能显著的提高微博信息推荐的准确性。
Context information is an important factor that affects people's interest. Based on the traditional LDA model, the context information is added to adjust the results of the LDA model. On the basis of the document-topic and topic-word, which is generated by the LDA model,the user interests is divided according to the different context information to generate the theme-mood distribution. On the basis of this,proposes a Time-LDA algorithm based on time context and a Mood-LDA algorithm based on the mood situation. Experiments on real data sets show that the algorithm proposed can significantly improve the accuracy of microblog information recommendation.