Spectral Subspace Clustering for Graphs with Feature Vectors

Günnemann, Stephan; Färber, Ines; Raubach, Sebastian; Seidl, Thomas

doi:10.1109/icdm.2013.110

Cited by 48 publications

(46 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Actually, the PFCD algorithm without nodal features considers community as dense subgraphs like most of other structure-based methods. The obtained results by considering just the structural properties are not good enough (less than 0.2) on certain type of networks such as DBLP, Arxiv, Internet, and Patent due to the fact that the community structures consist of the dense sub-graphs and assortative modules [49]. In addition, structure based methods perform better than the Plain on some networks such as Predator and PolBlogs (Figure 9) due to having small number of features.…”

Section: Resultsmentioning

confidence: 94%

Detection of Community Structures in Networks With Nodal Features based on Generative Probabilistic Approach

Zare

Hajiabadi

Jalili

2021

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Community detection is considered as a fundamental task in analyzing social networks. Even though many techniques have been proposed for community detection, most of them are based exclusively on the connectivity structures. However, there are node features in real networks, such as gender types in social networks, feeding behavior in ecological networks, and location on e-trading networks, that can be further leveraged with the network structure to attain more accurate community detection methods. We propose a novel probabilistic graphical model to detect communities by taking into account both network structure and nodes' features. The proposed approach learns the relevant features of communities through a generative probabilistic model without any prior assumption on the communities. Furthermore, the model is capable of determining the strength of node features and structural elements of the networks on shaping the communities. The effectiveness of the proposed approach over the state-of-the-art algorithms is revealed on synthetic and benchmark networks.

show abstract

Section: Resultsmentioning

confidence: 94%

Detection of Community Structures in Networks With Nodal Features based on Generative Probabilistic Approach

Zare

Hajiabadi

Jalili

2021

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

show abstract

“…Building on this intuition, various principles have been introduced, ranging from adaptations of hierarchical and spectral clustering [10,13,33], over block modeling [38] and generative models [40], to information theoretic principles [19,31] and the detection of quasi-cliques in node-labeled graphs [12]. We kindly refer to the excellent survey of [9] for a thorough discussion of community detection methods.…”

Section: Static Community Detectionmentioning

confidence: 99%

Discovery of “comet” communities in temporal and labeled graphs Com $$^2$$ 2

et al. 2015

View full text Add to dashboard Cite

While the analysis of unlabeled networks has been studied extensively in the past, finding patterns in different kinds of labeled graphs is still an open challenge. Given a large edge-labeled network, e.g., a time-evolving network, how can we find interesting patterns? We propose Com 2 , a novel, fast and incremental tensor analysis approach which can discover communities appearing over subsets of the labels. The method is (a) scalable, being linear on the input size, (b) general, (c) needs no user-defined parameters and (d) effective, returning results that agree with intuition. We apply our method to real datasets, including a phone call network, a computer-traffic network and a flight information network. The phone call network consists of 4 million mobile users, with 51 million edges (phone calls), over 14 days, while the flights dataset consists of 7733 airports and 5995 airline companies flying 67,663 different routes. We show that Com 2 spots intuitive patterns regarding edge labels that carry temporal or other discrete information. Our findings include large "star"-like patterns, near-bipartite cores, as well as tiny groups (five users), calling each other hundreds of times within a few days. We also show that we are able to automatically identify competing airline companies.

show abstract

“…The network has 26,114 vertices and 108,550 edges. • ARXIV [11]: This network is a citation network whose vertices represent papers and edges represent citation relationships. Attributes denote how often a specific keyword appears in the abstract of the paper.…”

Section: Real-world Graphsmentioning

confidence: 99%

“…We give the average scores over 50 runs in Table 2 and Table 3, each time with a randomly sampled vertex as the seed vertex. For SG-Pursuit, UNCut and AGC, we give them the same number of clusters as used in [11,47]. For each seed vertex, we first decide which cluster contains it and then compute the scores of these measures.…”

Section: Real-world Graphsmentioning

confidence: 99%

See 1 more Smart Citation

Incorporating User's Preference into Attributed Graph Clustering

Mautz

Böhm

et al. 2021

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Graph clustering has been studied extensively on both plain graphs and attributed graphs. However, all these methods need to partition the whole graph to find cluster structures. Sometimes, based on domain knowledge, people may have information about a specific target region in the graph and only want to find a single cluster concentrated on this local region. Such a task is called local clustering. In contrast to global clustering, local clustering aims to find only one cluster that is concentrating on the given seed vertex (and also on the designated attributes for attributed graphs). Currently, very few methods can deal with this kind of task. To this end, we propose two quality measures for a local cluster: Graph Unimodality (GU) and Attribute Unimodality (AU). The former measures the homogeneity of the graph structure while the latter measures the homogeneity of the subspace that is composed of the designated attributes. We call their linear combination as Compactness. Further, we propose LOCLU to optimize the Compactness score. The local cluster detected by LOCLU concentrates on the region of interest, provides efficient information flow in the graph and exhibits a unimodal data distribution in the subspace of the designated attributes.

show abstract

Spectral Subspace Clustering for Graphs with Feature Vectors

Cited by 48 publications

References 16 publications

Detection of Community Structures in Networks With Nodal Features based on Generative Probabilistic Approach

Detection of Community Structures in Networks With Nodal Features based on Generative Probabilistic Approach

Discovery of “comet” communities in temporal and labeled graphs Com $$^2$$ 2

Incorporating User's Preference into Attributed Graph Clustering

Contact Info

Product

Resources

About