社交网络中基于地理位置特征的社团发现方法研究与实现

发布时间：2018-11-15 13:49

【摘要】：随着内置定位芯片智能手机的广泛流行，促使传统的社交网向基于位置的社交网络发展。基于位置的社交网络是位置服务、移动互联网、以及传统的社交网络结合的产物，有更广泛的应用场景。用户在使用基于位置的社交网络时，会产生大量的含有地理位置信息的数据，如何利用这些由用户产生的大量的含有地理位置信息的数据，，分析用户的行为模式、运动轨迹、以及位置感知的用户社团结构等成为了研究热点。本文针对从新浪微博爬取的用户数据，首先分析了用户的地理位置特征，然后利用地理位置特征进行用户相似性计算和重叠社团的发现。提出了一种基于地理位置特征的用户相似度计算算法和一种基于地理位置特征的重叠社团发现算法，并设计实现了基于地理位置特征的重叠社团发现的可视化工具。论文的具体工作如下：研究了在基于位置的社交网中用户相似度的计算方法，通过分析用户的含有地理位置数据的特征，提出一种基于地理位置特征的用户相似度计算方法，并利用从新浪微博中爬取的数据验证了算法的有效性。研究了基于位置的社交网络中重叠社团的发现算法，在分析了社交网络中用户关系以及用户的地理位置特征基础上，改进边聚类算法，设计并实现了基于地理位置特征的重叠社团发现算法。最后通过实验证明了算法的有效性。本文设计并实现了基于地理位置特征的重叠社团发现的可视化工具。工具主要分为三层：数据层、核心层和视图层。数据层完成对用户签到记录数据的封装以及预处理；视图层主要完成显示功能，并触发相应的事件；核心层主要是算法的实现，包含主题提取模块、相似度计算模块、重叠社团发现模块，同时该层并负责对视图层产生的事件的做出响应和处理。
[Abstract]:With the popularity of smart phones with built-in locator chips, the traditional social networks are becoming more and more location-based. Location-based social networks are the combination of location services, mobile Internet, and traditional social networks. When users use location-based social networks, they will produce a large amount of data with geographical location information. How to use these data generated by users to analyze the behavior patterns and motion trajectories of users. And location-aware user community structure has become a research hotspot. Based on the user data crawled from Weibo of Sina, this paper first analyzes the geographical location features of users, and then calculates the similarity of users and the discovery of overlapping communities by using geographical location features. A user similarity calculation algorithm based on geographical location feature and an overlapping community discovery algorithm based on geographical location feature are proposed. The visualization tool of overlapping community discovery based on geographical location feature is designed and implemented. The main work of this paper is as follows: the method of calculating user similarity in location-based social network is studied, and the features of users with geographical location data are analyzed. A user similarity calculation method based on geographical location features is proposed, and the validity of the algorithm is verified by crawling data from Sina Weibo. In this paper, the algorithm of discovering overlapping communities in location-based social networks is studied. Based on the analysis of user relationships and geographical location characteristics of users in social networks, the edge clustering algorithm is improved. An overlapping community discovery algorithm based on geographical location features is designed and implemented. Finally, the effectiveness of the algorithm is proved by experiments. This paper designs and implements a visualization tool of overlapping community discovery based on geographical location features. Tools are divided into three layers: data layer, core layer and view layer. The data layer completes the encapsulation and preprocessing of the user check-in record data, the view layer mainly completes the display function and triggers the corresponding events. The core layer is mainly the implementation of the algorithm, including topic extraction module, similarity calculation module, overlapping community discovery module, and the layer is responsible for the response and processing of the events generated by the view layer.
【学位授予单位】：北京航空航天大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】