移动终端数据的社交环境挖掘

发布时间：2018-03-27 23:42

本文选题：移动设备　切入点：数据挖掘　出处：《吉林大学》2017年硕士论文

【摘要】：过去十年内,移动设备的发展非常迅速,尤其是近几年来智能手机的发展尤为迅猛。移动设备变得越来越小巧便携的同时,其功能却在不断增多,越来越多的移动设备及移动设备上的应用程序被开发出来。外形的便携和功能的丰富使得移动设备得到迅速的普及,极大地提高了移动设备的用户粘性,越来越多的人会随身携带移动设备,甚至人们的生活与之密不可分。随着移动设备上所集成的传感器元件的增多,我们可以随时随地搜集用户当前环境中不同种类的数据,为我们识别用户当前环境提供了丰富的信息。用户与移动设备之间的联系愈发紧密。尤其是当移动端的各类应用程序数量剧增的时候,人们在生活中也更加依赖于这些应用所提供的便捷服务。目前智能手机上大部分的应用软件所提供的功能和服务与用户当前的状态和环境无关,然而一个好的应用软件,尤其是与用户环境相关的应用,应该通过相关环境数据的分析和挖掘,了解或推断出该用户当前所在的位置信息以及用户当前的状态信息,这些信息对于支持应用制定用户个性化服务极为重要。个性化的设计能为用户带来愉悦的用户体验,例如为不同环境中的用户提供不同的服务;同时,移动端的数据挖掘可以帮助我们掌握大规模群体事件中的人群移动信息,有助于我们了解人类个体和群体的行为模式;移动端的数据还能够刻画或建模人们的活动行为、精确地预测人们的移动行为,在社会学、经济学、城市规划学等领域有着重要作用。因此,本工作利用移动设备端的传感器搜集用户的相关数据,通过对数据的分析和数据挖掘工作,对使用移动设备的用户识别或感知其环境、场景。本工作的主要工作难点如下:1.传统的移动设备对于用户的位置信息的识别依靠于移动设备上的传感器数据,这种通过传感器表达真实世界的方式往往具有很大的弊端:有时表达的不够准确;有时表达粒度过细;而且常常无法理解用户的需求。2.在相关的不同工作中对于“环境”的定义不同且不统一。某些文章中认为“环境”是指用户所在的地理位置,而某些文章则认为“环境”指用户当前的动作行为。3.在相关的“用户环境探测”或“环境感知”的相关工作中,分别提出了若干有效算法,然而却缺乏一个系统的框架。而本文的主要贡献在于:1.定义了问题相关的概念,包括用户属性、环境等;2.设计了一个可扩展的三层模型作为系统的框架,底层为传感器层,中间为属性层,顶层为环境层;3.设计了各层之间的映射算法,利用数据挖掘的方式探测和识别用户环境。
[Abstract]:Mobile devices have grown very rapidly in the past decade, especially in recent years. As mobile devices become smaller and more portable, their capabilities continue to grow. More and more mobile devices and applications on mobile devices have been developed. The convenience of appearance and the rich functions of mobile devices make mobile devices rapidly popular and greatly improve the user stickiness of mobile devices. More and more people are carrying mobile devices with them, and even people's lives are inseparable. With the increasing number of sensor components integrated into mobile devices, we can collect different kinds of data from users' current environment anytime and anywhere. It provides a wealth of information for us to identify the user's current environment. The connection between the user and the mobile device is getting closer, especially as the number of applications on the mobile side increases dramatically. People also rely more on the convenience of these apps in their lives. Most apps on smartphones now offer functions and services that are independent of the user's current state and environment, but a good application. In particular, applications related to the user environment should know or deduce the current location information of the user and the current status information of the user through the analysis and mining of the relevant environment data. This information is essential to support application development of user personalization services. Personalized design can provide users with a pleasant user experience, such as providing different services to users in different environments; at the same time, Data mining on mobile side can help us to grasp the information of crowd movement in large-scale group events, and help us to understand the behavior patterns of human individuals and groups, and the data of mobile side can also depict or model the behavior of people. Accurate prediction of people's mobile behavior plays an important role in sociology, economics, urban planning, and so on. Through data analysis and data mining, users using mobile devices are identified or aware of their environment, Scene. The main difficulties of this work are as follows: 1.The traditional mobile devices depend on the sensor data of the mobile device for the identification of the user's position information. This way of expressing the real world through sensors often has many disadvantages: sometimes it is not accurate enough, sometimes it is too granular. The definition of "environment" is different and inconsistent in related work. In some articles, "environment" refers to the geographical location of the user. Some articles think that "environment" refers to the current behavior of the user. 3. In the related work of "user environment detection" or "environment perception", several effective algorithms are put forward respectively. However, there is a lack of a systematic framework. However, the main contribution of this paper lies in the definition of problem-related concepts, including user attributes, environment, etc. 2. An extensible three-layer model is designed as the framework of the system, and the bottom layer is the sensor layer. The middle layer is the attribute layer, and the top layer is the environment layer. 3. The mapping algorithm between the layers is designed to detect and identify the user environment by the way of data mining.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212;TP311.13

【相似文献】