基于位置的社交网络用户轨迹相似性算法

发布时间：2019-04-17 12:43

【摘要】：为了提高在位置社交网络中匹配轨迹相似用户的准确性,提出了一种基于自适应密度聚类的用户轨迹相似性双重加权模型。首先,根据签到点的分布,提出基于聚类区域半径的自适应密度聚类算法,对签到点进行聚类,得到符合一定区域半径的自适应分层聚类区域。然后,针对用户相似性的计算,提出用户轨迹相似性双重加权模型:基于不同层次上权重不同及同一层次上不同签到区域权重不同2个原则。基于分层聚类区域,计算用户访问各分层的轨迹相似性;由于不同签到区域对相似度的表征能力不同,通过不同权重来计算具体层次上的相似性。进而,进行层次上的加权,获得用户总体的相似性。仿真表明,该方法能有效地分析出用户轨迹相似性,具备较高的准确性。
[Abstract]:In order to improve the accuracy of matching track similarity users in position social networks, a double weighted model based on adaptive density clustering is proposed. Firstly, according to the distribution of check-in points, an adaptive density clustering algorithm based on the radius of cluster region is proposed to cluster the check-in points and obtain the adaptive hierarchical clustering region which accords with a certain radius of the region. Then, according to the calculation of user similarity, a double weighted model of user track similarity is proposed, which is based on two principles: different weights at different levels and different check-in regions at the same level. Based on hierarchical clustering region, the similarity of users' access to each layer is calculated, and because different check-in regions have different representation ability to similarity, the similarity at specific levels is calculated by different weights. Then, we can get the similarity of the user population by weighting at the level. Simulation results show that this method can effectively analyze the similarity of user tracks and has high accuracy.
【作者单位】：四川大学电子信息学院;西南民族大学计算机学院;
【分类号】：TP393.09

【相似文献】