基于动态情境感知的W5模型研究
发布时间：2019-06-29 09:29
【摘要】：Twitter、Sina Micro-blog等社交网络应用为基于位置的服务提供了大量的情境信息,如用户ID(who)、签到时间(when)、GPS坐标(where)、微博内容主题词(what)和微博内容诱因词(why)等,简称5W。它们为用户的行为和偏好研究提供了契机。该文提出了基于5W动态情境感知信息的W5概率模型,并采用包含情境信息的联合概率分布分别从时间、空间和活动等方面挖掘用户动态行为,用于用户和位置的预测。该文实验基于两个数据集:Geo-text(GT)和Sina-tweets(ST),在数据集上进行了用户预测(UP)和位置预测(LP)实验。实验结果表明,W5模型在UP和LP两方面准确率均高于W4模型。同时,W5模型在时间误差和空间距离误差两方面也取得了较好的性能。
[Abstract]:Social network applications such as Twitter,Sina Micro-blog provide a lot of situational information for location-based services, such as user ID (who), check-in time (when), GPS coordinate (where), Weibo content theme (what) and Weibo content inducer (why), etc., abbreviated as 5W. They provide an opportunity for the study of users' behavior and preferences. In this paper, a W5 probability model based on 5W dynamic situational perception information is proposed, and the joint probability distribution containing situational information is used to mine the dynamic behavior of users from the aspects of time, space and activity respectively, which can be used to predict the user and location. In this paper, the experiment is based on two datasets: Geo-text (GT) and Sina-tweets (ST),. User prediction (UP) and location prediction (LP) experiments are carried out on the dataset. The experimental results show that the accuracy of W5 model is higher than that of W4 model in both UP and LP. At the same time, the W5 model also achieves good performance in terms of time error and spatial distance error.
【作者单位】： 武汉大学计算机学院;长江大学计算机科学学院;
【基金】：国家自然科学基金(61272109)
【分类号】：TP393.092
，
本文编号：2507717
[Abstract]:Social network applications such as Twitter,Sina Micro-blog provide a lot of situational information for location-based services, such as user ID (who), check-in time (when), GPS coordinate (where), Weibo content theme (what) and Weibo content inducer (why), etc., abbreviated as 5W. They provide an opportunity for the study of users' behavior and preferences. In this paper, a W5 probability model based on 5W dynamic situational perception information is proposed, and the joint probability distribution containing situational information is used to mine the dynamic behavior of users from the aspects of time, space and activity respectively, which can be used to predict the user and location. In this paper, the experiment is based on two datasets: Geo-text (GT) and Sina-tweets (ST),. User prediction (UP) and location prediction (LP) experiments are carried out on the dataset. The experimental results show that the accuracy of W5 model is higher than that of W4 model in both UP and LP. At the same time, the W5 model also achieves good performance in terms of time error and spatial distance error.
【作者单位】： 武汉大学计算机学院;长江大学计算机科学学院;
【基金】：国家自然科学基金(61272109)
【分类号】：TP393.092
，
本文编号：2507717
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2507717.html
