Mutually Enhancing Community Detection and Sentiment Analysis on Twitter Networks

The burgeoning use of Web 2.0-powered social media in recent years has inspired numerous studies on the content and composition of online social networks (OSNs). Many methods of harvesting useful information from social networks’ immense amounts of user-generated data have been successfully applied to such real-world topics as politics and marketing, to name just a few. This study presents a novel twist on two popular techniques for studying OSNs: community detection and sentiment analysis. Using sentiment classification to enhance community detection and community partitions to permit more in-depth analysis of sentiment data, these two techniques are brought together to analyze four networks from the Twitter OSN. The Twitter networks used for this study are extracted from four accounts related to Microsoft Corporation, and together encompass more than 60,000 users and 2 million tweets collected over a period of 32 days. By combining community detection and sentiment analysis, modularity values were increased for the community partitions detected in three of the four networks studied. Furthermore, data collected during the community detection process enabled more granular, community-level sentiment analysis on a specific topic referenced by users in the dataset.

Local Community Detection Using Link Similarity

Exploring local community structure is an appealing problem that has drawn much recent attention in the area of social network analysis. As the complete information of network is often difficult to obtain, such as networks of web pages, research papers and Facebook users, people can only detect community structure from a certain source vertex with limited knowledge of the entire graph. The existing approaches do well in measuring the community quality, but they are largely dependent on source vertex and putting too strict policy in agglomerating new vertices. Moreover, they have predefined parameters which are difficult to obtain. This paper proposes a method to find local community structure by analyzing link similarity between the community and the vertex. Inspired by the fact that elements in the same community are more likely to share common links, we explore community structure heuristically by giving priority to vertices which have a high link similarity with the community. A three-phase process is also used for the sake of improving quality of community structure. Experimental results prove that our method performs effectively not only in computer-generated graphs but also in real-world graphs.

Enhancing Sentiment Analysis on Twitter Using Community Detection

The increasing popularity of social media in recent years has created new opportunities to study the interactions of different groups of people. Never before have so many data about such a large number of individuals been readily available for analysis. Two popular topics in the study of social networks are community detection and sentiment analysis. Community detection seeks to find groups of associated individuals within networks, and sentiment analysis attempts to determine how individuals are feeling. While these are generally treated as separate issues, this study takes an integrative approach and uses community detection output to enable community-level sentiment analysis. Community detection is performed using the Walktrap algorithm on a network of Twitter users associated with Microsoft Corporation’s @technet account. This Twitter account is one of several used by Microsoft Corporation primarily for communicating with information technology professionals. Once community detection is finished, sentiment in the tweets produced by each of the communities detected in this network is analyzed based on word sentiment scores from the well-known SentiWordNet lexicon. The combination of sentiment analysis with community detection permits multilevel exploration of sentiment information within the @technet network, and demonstrates the power of combining these two techniques.

Advanced Community Identification Model for Social Networks

Community detection in social networks is a hard problem because of the size,and the need of a deep understanding of network structure and functions.While several methods with significant effort in this direction have been devised,an outstanding open problem is the unknown number of communities,it is generally believed that the role of influential nodes that are surrounded by neighbors is very important.In addition,the similarity among nodes inside the same cluster is greater than among nodes from other clusters.Lately,the global and local methods of community detection have been getting more attention.Therefore,in this study,we propose an advanced communitydetection model for social networks in order to identify network communities based on global and local information.Our proposed model initially detects the most influential nodes by using an Eigen score then performs local expansion powered by label propagation.This process is conducted with the same color till nodes reach maximum similarity.Finally,the communities are formed,and a clear community graph is displayed to the user.Our proposed model is completely parameter-free,and therefore,no prior information is required,such as the number of communities,etc.We perform simulations and experiments using well-known synthetic and real network benchmarks,and compare them with well-known state-of-the-art models.The results prove that our model is efficient in all aspects,because it quickly identifies communities in the network.Moreover,it can easily be used for friendship recommendations or in business recommendation systems.

Social Opinion Network Analytics in Community Based Customer Churn Prediction

Community based churn prediction,or the assignment of recognising the influence of a customer’s community in churn prediction has become an important concern for firms in many different industries.While churn prediction until recent times have focused only on transactional dataset(targeted approach),the untargeted approach through product advisement,digital marketing and expressions in customer’s opinion on the social media like Twitter,have not been fully harnessed.Although this data source has become an important influencing factor with lasting impact on churn management.Since Social Network Analysis(SNA)has become a blended approach for churn prediction and management in modern era,customers residing online predominantly and collectively decide and determines the momentum of churn prediction,retention and decision support.In existing SNA approaches,customers are classified as churner or non-churner(1 or 0).Oftentimes,the customer’s opinion is also neglected and the network structure of community members are not exploited.Consequently,the pattern and influential abilities of customers’opinion on relative members of the community are not analysed.Thus,the research developed a Churn Service Information Graph(CSIG)to define a quadruple churn category(churner,potential churner,inertia customer,premium customer)for non-opinionated customers via the power of relative affinity around opinionated customers on a direct node to node SNA.The essence is to use data mining technique to investigate the patterns of opinion between people in a network or group.Consequently,every member of the online social network community is dynamically classified into a churn category for an improved targeted customer acquisition,retention and/or decision supports in churn management.

A case study on knowledge communication based on friends-list links in the science blogging community at Sciencenet.cn

Sciencenet.cn is the leading online portal serving the Chinese scientific community. This paper intends to analyze the interdisciplinary and intradisciplinary knowledge communication patterns based on friends-list links in the blog community at Sciencenet.cn by using hyperlink analysis and social network analysis. The major findings are: 1) More bloggers have an academic background in management science and life science; 2) there are some core actors in co-inlink network and co-outlink network, who take the lead in engaging with knowledge exchange activities and produce a great influence on interdisciplinary communication; 3) interactive relationships commonly exist between a blogger and those on his/her friends list, and the most linked-to blogs usually play a key role in generating interactive communication; 4) management science has the highest co-inlink count with life science or information science and it has the highest co-outlink count with life science or mathematical and physical science; 5) management science and life science have the greatest impact on information science and the interdisciplinary knowledge communication will also produce relatively significant influence on the development of information science discipline. It is our hope that this research can serve as a reference source for the future studies of academic virtual communities, and the development of mechanisms for facilitating increased engagement in knowledge exchange activities in academic virtual communities.

Community Detection in Complex Networks

With the rapidly growing evidence that various systems in nature and society can be modeled as complex networks, community detection in networks becomes a hot research topic in physics sociology, computer society, etc. Although this investigation of community structures has motivated many diverse algorithms, most of them are unsuitable when dealing with large networks due to their computational cost. In this paper, we present a faster algorithm ComTector which is more efficient for the community detection in large complex networks based on the nature of overlapping cliques. This algorithm does not require any priori knowledge about the number or the original division of the communities. With respect to practical applications, ComTector is challenging with five different types of networks including the classic Zachary Karate Club, Scientific Collaboration Network South Florida Free Word Association Network, Urban Traffic Network North America Power Grid and the Telecommunication Call Network. Experimental results show that our algorithm can discover meaningful communities that meet both the objective basis and our intuitions.

SSDBA:the stretch shrink distance based algorithm for link prediction in social networks

In the field of social network analysis,Link Predic-tion is one of the hottest topics which has been attracted attentions in academia and industry.So far,literatures for solving link prediction can be roughly divided into two categories:similarity-based and learning-based methods.The learning-based methods have higher accuracy,but their time complexities are too high for complex networks.However,the similarity-based methods have the advantage of low time consumption,so improving their accuracy becomes a key issue.In this paper,we employ community structures of social networks to improve the prediction accuracy and propose the stretch shrink distance based algorithm(SSDBA),In SSDBA,we first detect communities of a social network and identify active nodes based on community average threshold(CAT)and node average threshold(NAT)in each community.Second,we propose the stretch shrink distance(SSD)model to iteratively calculate the changes of distances between active nodes and their local neighbors.Finally,we make predictions when these links'distances tend to converge.Furthermore,extensive parameters learning have been carried out in experiments.We compare our SSDBA with other popular approaches.Experimental results validate the effectiveness and efficiency of proposed algorithm.

基于链路预测的社会网络事件检测方法

网络演化分析与事件检测,是当前社会网络研究的热点和难点.现有的研究工作主要是针对网络提出不同的模型,并用网络特征指标对仿真结果进行评价.这些方法存在如下问题:(1)每种方法仅针对特定网络,通用性不高;(2)特征指标多种多样,不同模型的表现情况缺乏统一的评价标准;(3)未考虑网络演化的时间特性,难以描述网络演化的波动性,无法检测事件.针对上述问题,提出一种基于链路预测的社会网络事件检测方法 Link Event(由相似性计算算法Sim C和事件检测算法Event D组成),它可以对不同网络的波动性进行统一评价,并依此建立事件检测模型.主要工作包括:(1)证明了链路预测可以反映网络演化机制,相同机制下的模型演化法和链路预测在分析网络演化上具有内在的一致性;(2)基于链路预测,提出一种网络相似性计算算法Sim C(similar computing),并在考虑微观因素的基础上进行改进;(3)利用相似性计算结果,提出一种事件检测算法Event D(event detecting)检测出新事件.在不同特征的网络上进行实验,结果表明:所提出的Link Event方法能够较好地解决网络演化波动性问题,实现事件检测;同时也证明了利用链路预测技术进行网络演化分析的可行性以及相似性计算和事件检测算法的有效性.

校园无线网用户群体的移动行为聚集分析

寻找更好更高效的计算用户之间相似度的方法是个难题,聚集结果对网络运维的帮助也较少被关注。提出了终端移动轨迹的稀疏链接区间(SLI,sparse linked intervals)概念,以此为基础使用社会网络分析的方法有效地分析了移动终端的聚集关系。主要采用了北京大学无线校园网真实的实际运行数据进行分析,并用公开数据集进行了验证。实验结果表明,提出的方法能够很好地发现用户群体。还分析了3种常见的聚集层次子图模式,以及它们的形成原因和与无线网络管理的联系。

复杂网络半监督的社区发现算法研究

为提高社区发现算法的运行效率,提出了一种基于节点相似度的半监督社区发现算法——SSGN算法。充分利用先验知识must-link、cannot-link约束集合,将先验信息通过衍生规则进行扩展,并对扩展的信息通过基于距离度量的方式加以验证。采用人工网络在UCI数据集和大型真实数据集上与真实网络进行验证,实验结果表明,基于节点相似度的半监督社区发现算法较其他半监督聚类算法更准确,也更高效。

社交网络的舆情热点发现模型研究

随着社交网络的大规模发展,传统的基于文本分析的热点话题发现技术面临严峻的挑战。构建了针对社交媒体的舆情热点发现分层架构模型,详细介绍了模型中的关键技术,并将时间因素和链接分析技术引入舆情热点发现算法,阐述了新算法的主要思想和大致流程。在舆情热点发现中引入网络化方法,开拓了舆情分析的新思路。

基于形式概念分析的博客社区发现

针对拖网算法存在的发现Web社区数量过多、社区间页面重复率较高以及严格的社区定义形成孤立社区等问题,提出一种基于形式概念分析(FCA)的博客社区发现算法。根据博客网络之间的链接关系构造概念格,通过格的代数消解对原始概念格进行等价划分,度量每个划分中概念间外延和内涵的结构相似性进而合并社区核心形成社区。实验结果表明:测试数据集中社区核心的网络密度大于40%的占全部的83.420%,合并社区的网络直径为3,且社区内容丰富程度得到提高。所提算法可以有效地运用于博客、微博等社交网络的社区发现,具有显著的应用价值和现实意义。

一种采用社团信息的链接预测方法

链接预测研究如何利用网络中已有的信息预测可能存在的关系链接,目前已成为数据挖掘领域的热点研究问题之一。社会网络中普遍存在社团结构,社团对链接的形成有重要的影响,但在大多数链接预测方法中未得到深入研究。针对这一现象提出一种新的链接预测方法,采用社团信息改进节点对样本的描述,并在监督学习框架中学习和预测。在现实数据集Facebook和ACF中的实验结果表明,加入社团信息的链接预测方法获得了更高的准确率。

Exploratory analysis of grocery product networks

Finding meaningful sets of co-purchased products allows retailers to manageinventory better and develop market strategies. Analyzing the baskets ofproducts, known as market basket analysis, is typically carried out usingassociation rule mining or community detection approach. This article usesboth methods to investigate a transaction dataset collected from a brick-andmortargrocery store. The findings reveal interesting purchasing patterns oflocal residents and prompt us to consider dynamic modeling of the productnetwork in the future.