日常交互中朋友关系强度度量方法

发布时间：2018-05-18 03:36

本文选题：关系强度 + 轨迹相似度　；参考：《国防科技大学学报》2017年03期

【摘要】：针对如何度量日常生活中人们之间的关系强度问题展开研究,提出一个从日常轨迹、语义位置以及语义标签三个层次度量朋友之间关系强度的层级模型FRSHV。采用动态时间规整模型通过计算朋友之间的空间距离来度量其日常轨迹之间的相似度,进而使用轨迹序列熵值对用户每天轨迹的相似度进行加权处理,将其作为朋友之间的关系强度;采用主题模型隐含狄利克雷分布分别计算朋友之间的基于语义位置和语义标签的行为模式的相似性,将其作为朋友之间的关系强度;采用集成学习的思想对三个层次的度量结果进行投票,以投票结果作为最终的朋友之间的关系强度。在公开数据集上对FRSHV模型的有效性进行实验验证,结果表明该模型能够有效地度量朋友之间的关系强度。
[Abstract]:Based on the study of how to measure the relationship strength between people in daily life, a hierarchical model named FRSHV is proposed to measure the relationship strength between friends from three levels: daily locus, semantic location and semantic label. The dynamic time warping model is used to measure the similarity between the daily trajectories of friends by calculating the space distance between them, and then the similarity of the user's daily trajectory is weighted by using the entropy value of the trajectory sequence. It is regarded as the strength of the relationship between friends, and the similarity of the behavior patterns based on semantic location and semantic label is calculated by using the implicit Delikley distribution of the topic model, which is regarded as the strength of the relationship between friends. The idea of integrated learning is used to vote the measurement results at three levels, and the voting results are used as the ultimate relationship between friends. The validity of the FRSHV model is verified by experiments on the open data set. The results show that the model can effectively measure the relationship between friends.
【作者单位】：国防科技大学计算机学院;
【基金】：国家自然科学基金资助项目(61202117,91118008)
【分类号】：P228.4;TN929.5

【相似文献】