Comparaison des méthodes de classification pour l'identification des noeuds importants dans les graphes dynamiques

Abstract : Nowadays, researchers are interested in the detection of important entities in networks ; this can be keywords in a document and Twitter or more even super-spreaders in a movement network. One natural way to detect these entities is to use graph theory; each entity is represented by a node in the graph. Afterward, centrality metrics such as Temporal closeness can be applied to detect the important nodes. Nevertheless, this can be computationally expensive. In this work, we examine three basic characteristics that we consider as the basic blocks of Temporal closeness. We utilize those characteristics to show that classifiers are capable to classify the nodes. In addition, we show that taking into account the dataset's nature does not necessarily produce better models. Finally, we compare the computational time of these models against that of Temporal Closeness.
Complete list of metadatas

Cited literature [21 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02085267
Contributor : Lionel Tabourier <>
Submitted on : Saturday, March 30, 2019 - 4:29:35 PM
Last modification on : Friday, July 5, 2019 - 3:26:03 PM

File

example1.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02085267, version 1

Citation

Marwan Ghanem. Comparaison des méthodes de classification pour l'identification des noeuds importants dans les graphes dynamiques. Rencontres jeunes chercheurs en RI, Mar 2019, Lyon, France. ⟨hal-02085267⟩

Share

Metrics

Record views

40

Files downloads

51