摘要
文章提出了与时间因素相结合的特征词权重更新方式以及关联分析与TF-IDF相结合的特征词推荐策略,通过实验数据证明可以及时更新特征词权重、大幅提高特定话题相关特征词推荐的有效性,从而提高对用户分类的准确度。文章提出的两个策略优于传统策略,通过实验数据证明有效可行。
The solutions presented in this article isthat update the weight with time factor and the new feature words recommendation strategy combine with TF-IDF. The data bear out that it can update the weight timely and improve the microblogusers classification,and the high degreeofaccurancy in feature words recommendation.The two strategies presented in this paper do well in the actual word, and the data bear out that two strategies is feasible.
出处
《信息网络安全》
2012年第8期136-139,共4页
Netinfo Security
基金
国家863计划项目[2010AA012505
2011AA010702
2012AA01A401
2012AA01A402]
自然科学基金课题[60933005]
科技支撑计划课题[2012BAH38B04]
国家242信息安全计划课题[2011A010]
