Éditeur(s) : HAL CCSD
Résumé : International audience
One of the most common tasks in the area of social network mining is the extraction of frequent patterns from social networks. Although traditional approaches have been mainly focused on subgraphs occurring frequently in a network or a set of networks, new approaches have attempted to exploit network structure and node properties in order to elicit new kinds of patterns. One of these news approaches is the extraction of conceptual links, a solution that combines both structure and node properties for providing knowledge on the groups of nodes the most connected in a social network. However, if the extraction of conceptual links offers a great potential in terms of knowledge discovery as well as network visualization, the search for these kinds of patterns remains a computationally intensive problem. Thus, the efficiency of the extraction processes highly depends on the efficiency of the underlying algorithm. In this paper, we focus on the extraction of maximal frequent conceptual links (MFCL) and we propose a hybrid algorithm that applies a filter on the nodes for reducing the search space to the most frequent groups of nodes. Although our solution potentially causes loss of some patterns, we demonstrate its efficiency by comparison with the MFCLMin algorithm. We investigate the potential loss of patterns, the gain on the runtime, and the gain on the number of comparisons.
IEEE International Conference on Research Challenges in Information Science
Paris, France

hal-00841699
https://hal.archives-ouvertes.fr/hal-00841699

Éditeur(s) : HAL CCSD
Résumé : International audience
In this work, we propose a novel approach for the discovery of frequent patterns in a social network on the basis of both vertex attributes and link frequency. With an analogy to the traditional task of mining frequent item sets, we show that the issue addressed can be formulated in terms of a conceptual analysis that elicits conceptual links. A social-based conceptual link is a synthetic representation of a set of links between groups of vertexes that share similar internal properties. We propose a first algorithm that optimizes the search into the concept lattice of conceptual links and extracts maximal frequent conceptual links. We study the performances of our solution and give experimental results obtained on a sample example. Finally we show that the set of conceptual links extracted provides a conceptual view of the social network.
IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM)
Istanbul, Turkey

hal-00786161
https://hal.archives-ouvertes.fr/hal-00786161
DOI : 10.1109/ASONAM.2012.15

Éditeur(s) : HAL CCSD
Résumé : International audience
In this article, we analyse a collaborative network to understand the underlying patterns that structure the co-writing process of scientific articles. Our goal is to identify and understand the collaboration tendencies from authors publishing activities. For this purpose, we adopt a descriptive modelling through a network approach that consists first in generating the collaborative network from data on publications. Nodes of the network are then enriched with a set of individual attributes extracted from the publishing activity of each author. Finally, we search for conceptual views, a recent link clustering approach, which allows to summarize any kind of networks by highlighting the sets of attributes found frequently linked. Results show that it exists strong tendencies that unconsciously structure the collaboration behaviours. In this paper, we present these tendencies and show how they evolve according to different extraction thresholds.
EEE/ACM International Conference on Advances in Social Networks Analysis and Mining
Sydney, Australia

hal-01637563
https://hal.archives-ouvertes.fr/hal-01637563
DOI : 10.1145/3110025.3110146

Éditeur(s) : HAL CCSD
Résumé : International audience
Numerous social network mining methods have been proposed until now for addressing social mining tasks and specially searching for communities or frequent social patterns. However, the degree of complementarity of these methods has been very little studied. In this paper, we focus on two knowledge extraction processes in social networks: a link-based clustering that may extract social communities and a recent approach that searches for frequent conceptual link involving both clustering and search for frequent social patterns. We explore how the models extracted by each method may match and which potential useful knowledge they may provide. Our objective is to evaluate the potential relationships between communities and frequent conceptual links. For this purpose, we propose a set of measures for evaluating the degree to which these patterns extracted from the same dataset are matching. Our approach is applied on two datasets and demonstrates the importance of considering simultaneously various kinds of knowledge and their complementarity.
IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining
Niagara falls, Canada

hal-00874351
https://hal.archives-ouvertes.fr/hal-00874351

Éditeur(s) : HAL CCSD
Résumé : International audience
The extraction of knowledge from social networks is an area that has experienced significant growth in recent years. Indeed, thanks to the improvement of storage and calculation capacities, and the heterogeneity of data that can currently be extracted, much effort has been made to go beyond traditional knowledge, by proposing new kinds of patterns that take into account the context. However, while many works were interested in designing new patterns of knowledge or in optimizing existing approaches, few studies have been focused in merging patterns and on the useful knowledge emerging from such fusions. In this work, we focus on two network clustering approaches, able to extract two distinct kinds of patterns, and we seek to understand both the intersections that can exist between them and the knowledge that emerges from their fusion. The first is the classical nodes clustering approach that consists in searching for communities into a network. The second is the search for frequent conceptual links, a new link clustering approach that aims identifying frequent links between groups of nodes sharing common attributes. We propose a set of original measures that aim to evaluate the amount of shared information between these patterns when they are extracted from a same network. These measures are applied to three datasets and demonstrate the interest in simultaneously considering several sources of knowledge.
Advances in Knowledge Discovery and Management (AKDM)

hal-01514227
https://hal.archives-ouvertes.fr/hal-01514227
DOI : 10.1007/978-3-319-45763-5_13

Éditeur(s) : HAL CCSD Springer
Résumé : International audience
he search for frequent conceptual links (FCL) is a new network clustering approach that exploits both the network structure and the node properties. While recent works have been focused on the optimisation of the FCL extraction, no study has been conducted on the matching or the intersections between the conceptual links and the classical clustering approach which consists in extracting communities. In this paper, we focus on these two approaches. Our objective is to evaluate the possible relationships existing between FCL and communities to understand how the patterns obtained by each kind of methods may match as well as the useful knowledge resulting from the intersections of these two kinds of knowledge. For this purpose, we propose a set of original measures, based on the notion of homogeneity, assessing the level of shared information between these patterns when they are extracted from the same dataset. The measures are applied on three datasets and demonstrate the importance of considering simultaneously various kinds of knowledge and their intersections.
ISSN: 1869-5450

hal-01312313
https://hal.archives-ouvertes.fr/hal-01312313
DOI : 10.1007/s13278-014-0197-1