高级检索

    马宇舸, 程华, 寇晓淮, 林家骏. 软信息的概率特征关联算法[J]. 华东理工大学学报(自然科学版), 2017, (1): 84-89. DOI: 10.14135/j.cnki.1006-3080.2017.01.014
    引用本文: 马宇舸, 程华, 寇晓淮, 林家骏. 软信息的概率特征关联算法[J]. 华东理工大学学报(自然科学版), 2017, (1): 84-89. DOI: 10.14135/j.cnki.1006-3080.2017.01.014
    MA Yu-ge, CHENG Hua, KOU Xiao-huai, LIN Jia-jun. Probabilistic Feature Association Algorithm of Soft Information[J]. Journal of East China University of Science and Technology, 2017, (1): 84-89. DOI: 10.14135/j.cnki.1006-3080.2017.01.014
    Citation: MA Yu-ge, CHENG Hua, KOU Xiao-huai, LIN Jia-jun. Probabilistic Feature Association Algorithm of Soft Information[J]. Journal of East China University of Science and Technology, 2017, (1): 84-89. DOI: 10.14135/j.cnki.1006-3080.2017.01.014

    软信息的概率特征关联算法

    Probabilistic Feature Association Algorithm of Soft Information

    • 摘要: 基于软信息的新闻事件态势估计中,事件当前态势的准确估计需融合事件的长期态势。以长期词典作为事件长期态势的特征表达,提出了软信息的长期特征与当前特征关联融合的概率特征关联算法。由事件的长期信息抽取得到长期词典,基于特征词相似度将长期特征与当前特征进行概率关联,得到事件的全特征表达,并提出了特征的长期关联度指标与类别关联度指标评估概率特征关联算法的有效性。实验结果表明,概率特征关联算法能够有效地融合长期态势,提升事件当前态势的估计精度。

       

      Abstract: The situation assessment based on news events should consider the long-term trend of the events.In this paper,the long-term dictionary is introduced to characterize the long-term trend,and then,a probabilistic feature association algorithm is proposed for long-term features and current features.In order to obtain the full feature of the news event,the proposed algorithm firstly extracts long-term dictionary based on long-term text information collection of a news event.Besides,the probabilistic feature association algorithm,which is based on the similar degree of the keywords,is utilized to fuse the long-term feature into the current feature.In order to evaluate the association algorithm performance,both long-term association degree and class association degree are proposed.The experimental results show that the probabilistic feature association algorithm can introduce the long-term trend and improve the accuracy of situation assessment.

       

    /

    返回文章
    返回