微博作为一种新兴的社交网络正在改变我们生活,对社会诸领域的介入和渗透与日俱增,现在许多民意调查通过对微博话题的舆情分析完成.针对现有微博研究中情感分析的对象较为单一问题,通过面向多维关联的数据进行情感分析,融合微博话题扩散度和话题情感倾向得到微博话题的舆情值,增强结果的有效性.微博话题的情感倾向分析将表情符号考虑在内,通过FP增长算法和检索距离建立表情符号库,运用朴素贝叶斯方法完成情感倾向分析.对新浪微博网络数据集进行实验,并以查全率与查准率为评估标准,结果表明该方法准确率可达82.6%. As a new social network, microblogging is changing our lives and increasing involvement and infiltration in all fields of society. Nowadays many public opinion surveys are completed through the public opinion analysis of Weibo topics. Aiming at the existing objects of the emotional analysis in the existing Weibo study Single problem, through multidimensional correlation of the data for emotional analysis, the integration of micro-blog topic diffusion degree and the topic of emotional orientation to get the value of the Weibo topic of public opinion, and enhance the effectiveness of the results.Weibo topic emotional propensity analysis will be emoji , The FP growth algorithm and the retrieval distance are used to build the emoticon library, and the naive Bayesian method is used to complete the analysis of the affective tendencies.Experiments are conducted on the Sina Weibo network dataset, and the recall rate and accuracy rate are used as evaluation criteria The accuracy of the method can reach 82.6%.