Data Portal @ linkeddatafragments.org

Matches in ScholarlyData for { <https://w3id.org/scholarlydata/inproceedings/www2008/paper/310> ?p ?o. }

• 310 creator belle-tseng.
• 310 creator hari-sundaram.
• 310 creator shenghuo-zhu.
• 310 creator yu-ru-lin.
• 310 creator yun-chi.
• 310 type InProceedings.
• 310 label "FacetNet: A Framework for Analyzing Communities and Their Evolutions in Dynamic Networks".
• 310 sameAs 310.
• 310 abstract "In many networked data, such as social networks and paper co-authorship networks, individuals form various communities through interactions. Very often, these communities evolve continuously with time, due to, for example, the change of a person's social status or the change of an author's research interest. Discovering communities and analyzing their evolutions are important research problems. Traditionally, these two problems are usually solved separately, i.e., through a two-stage approach where in the first stage communities are extracted for each timestep and in the second stage, the extracted communities of different timesteps are compared to study their evolutions. We argue that such a two-stage approach is inappropriate in many applications where data are often noisy and algorithms are not necessarily robust. In contrast, in this paper we propose a novel framework, FacetNet, for analyzing communities and their evolutions through a unified process. This framework will discover communities that both fit well with current observed networked data and evolve smoothly over time. We present our framework in the form of non-negative matrix factorization, while at the same time we also provide justification to our formulation from the information theory and probabilistic model points of view. To solve our formulation, we provide an iterative algorithm that is guaranteed to converge to an optimal solution. The algorithm has provable low time complexity. We use extensive experimental studies, both on synthetic datasets and on real datasets, to demonstrate that our method gives more reasonable communities and provides insights that are not directly obtainable from traditional methods.".
• 310 hasAuthorList authorList.
• 310 hasTopic World_Wide_Web.
• 310 isPartOf proceedings.
• 310 keyword "Community Net".
• 310 keyword "Community".
• 310 keyword "Evolution Net".
• 310 keyword "Evolution".
• 310 keyword "Soft Modularity".
• 310 title "FacetNet: A Framework for Analyzing Communities and Their Evolutions in Dynamic Networks".