Finding Overlapping Communities in Social Networks

Goldberg, Mark; Kelley, Stephen; Magdon‐Ismail, Malik; Mertsalov, Konstantin; Wallace, Al

doi:10.1109/socialcom.2010.24

Cited by 47 publications

(21 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We think deploying social sentiment to play a central role in producing recommendations is still rarely explored and therefore entails our work. Last but not least, community formation and detection is also closely related to the field of SNS interacting with recommendation systems ( [10], [11], and [23]). …”

Section: Data From Social Networkmentioning

confidence: 99%

Crowdsourcing recommendations from social sentiment

Xie

Honbo

Zhang

et al. 2012

Proceedings of the First International Workshop on Issues of Sentiment Discovery and Opinion Mining

View full text Add to dashboard Cite

In this paper, we investigate an innovative recommendation system by incorporating relevant social opinion and sentiment information. Our recommendation system, a powerful application of social sentiment analysis, differs from many existing models, which investigate the situation where the social network itself is structured to work with the product ranking and is specially built inside an e-commerce website. In contrast, our proposed system focuses on constructing and inferring product recommendations from external social network services (SNS) such as Facebook. In our system, we process product features in a finite-dimensional polynomial linear space. Additional components of our proposed system include an asymmetric similarity measurement and an asymmetric advantage measurement. We also show that our definitions for the two measurements include specific properties that reduce the computational overhead in the experiments. An important aspect of our modeling is to incorporate user-generated high-level semantic sentiment in the analysis. We apply our models to real time data and observe promising results for not only product recommendation but also job recommendation.

show abstract

Section: Data From Social Networkmentioning

confidence: 99%

Crowdsourcing recommendations from social sentiment

Xie

Honbo

Zhang

et al. 2012

Proceedings of the First International Workshop on Issues of Sentiment Discovery and Opinion Mining

View full text Add to dashboard Cite

show abstract

“…The static communities are then analyzed to detect evolutionary behavior. Specifically, the communities are discovered using LOS clustering [13] and evolutions with an F (·) value above 0.2 were taken to be valid. Both networks resulted in about 50,000 evolutions.…”

Section: Determining Communities and Evolutionsmentioning

confidence: 99%

Identifying Long Lived Social Communities Using Structural Properties

Goldberg

Magdon‐Ismail

Thompson

2012

2012 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining

View full text Add to dashboard Cite

Abstract-We present a two step procedure to identify long lasting communities, or evolutions, in social networks. First, we use axiomatic foundations to 'rigorously' establish shorter, strongly-connected evolutions. In the second step, we use heuristics to combine these shorter evolutions to form longer evolutions. We apply the procedure on data generated from two networksthe DBLP co-authorship database and LiveJournal blog data. We visually validate our algorithms by examining the topic evolution of the associated documents. Our results demonstrate that our algorithms, based solely on structural properties of the data (who interacts with whom), are able to track thematic trends in the literature. We then use a machine learning framework to identify the structural features of the early stages of a community's evolution are most useful for predicting the lifetime of the community. We find that (in order) size, intensity and stability are the most important features.

show abstract

“…To this end, we need a community fitness function in order to quantify the similarity between a node u and a neighbor community C. We find the fitness function F S = |S in | 2|S in |+|S out | (where S ⊆ V ) commonly used in [11] [14][17] performs competitively in both synthesized and real-world datasets.…”

Section: Revisiting Unassigned Nodesmentioning

confidence: 99%

“…Unlike the non-overlapping point of view, a (overlaped) community should be local and independent of its context or topological environment, as proposed in a recent work [10]. In addition, Goldberg [11] suggests that a community should also satisfy connectedness, i.e., it should induce a connected sub-graph in the network, and local optimality, i.e., the removal or addition of a single node will not affect the community with respect to a density function. While we agree upon the independence and connectedness properties, we find the local optimality too strict to allow the extension of network communities, especially when they can overlap.…”

Section: Introductionmentioning

confidence: 99%

Overlapping Community Structures and Their Detection on Social Networks

Nguyen

Dinh

Nguyen

et al. 2011

2011 IEEE Third Int'l Conference on Privacy, Security, Risk and Trust and 2011 IEEE Third Int'l Conference on Social Computing

View full text Add to dashboard Cite

Abstract-We propose DOCA (Detecting Overlapping Community Algorithm), a connection-based algorithm for discovering high quality overlapping community structures in social networks. Our proposed method is fast, very limited parameter dependent and only requires local knowledge about the network topology. Furthermore, the community structures discovered by DOCA are deterministic, i.e., no fuzzy community assignments are produced. DOCA's performance is certified by extensive experiments on real-world traces including Enron communication network, ArXiv citation and Astro physics collaboration networks as well as Facebook and Foursquare social networks. The demonstrative benchmark with other detection methods highlights the efficiency of DOCA when discovering community structures of large-scale networks. By using DOCA to analyze the community structures of real datasets, we find that overlapping communities occur naturally and quite frequently, especially for top largest communities. In addition, overlapped nodes tend to be active users who participate in multiple communities at the same time. This happens not only on social networks but also on collaboration, citation and communication networks.

show abstract

Finding Overlapping Communities in Social Networks

Cited by 47 publications

References 18 publications

Crowdsourcing recommendations from social sentiment

Crowdsourcing recommendations from social sentiment

Identifying Long Lived Social Communities Using Structural Properties

Overlapping Community Structures and Their Detection on Social Networks

Contact Info

Product

Resources

About