Cornet

These scripts are meant to extract certain features from raw Twitter data describing Twitter users (tweets, profile info, as well as external data). Once the features are extracted, various forms of SVMs are trained, and logistic regressions are performed, to classify and rank the users. These operations are conducted on different subgroups of features. The details of the process are given in the below publications. The scripts were applied to the classification/ranking of Twitter users in terms of offline influence, based on the RepLab 2014 dataset.

This software aims at studying social capitalists, which are a specific type of users of social networks services such as Twitter. The tool is generic, so it can actually be applied to completely different systems, as long as they can be represented as directed networks (i.e. digraphs). We applied our tool to Twitter in several research papers, listed below. Our work was also mentioned on the blog of the MIT Technology review. URL: https://github.com/CompNet/Orleans Release date: 2013–2014 Related publications: Nicolas Dugué, Vincent Labatut et Anthony Perez. « A Community Role ApproachTo Assess Social Capitalists Visibility in the Twitter Network ». In : Social NetworkAnalysis and Mining (SNAM) 5:26 (2015). DOI: 10.1007/s13278-015-0266-0.⟨hal-01163741⟩ Nicolas Dugué, Vincent Labatut et Anthony Perez. « Identifying the CommunityRoles of Social Capitalists in the Twitter Network ». In : IEEE/ACM InternationalConference on Advances in Social Network Analysis and Mining (ASONAM). Beijing,CN : IEEE Publishing, 2014, p. 371-374. DOI: 10.1109/ASONAM.2014.6921612.⟨hal-01011910⟩ (cite this publication if you use this software) Nicolas Dugué, Vincent Labatut et Anthony Perez. « Identification de rôles communautairesdans des réseaux orientés, appliquée à Twitter ». In : 14ème ConférenceExtraction et Gestion des Connaissances (EGC). Rennes, FR, 2014, p. 125-130. EGC ⟨hal-00918175⟩ Nicolas Dugué, Vincent Labatut et Plus d'infos

These scripts implement several measures allowing to compare two community structures, i.e. two partitions of the node set of a given graph. They are based on popular measures defined in the field of cluster analysis, namely: The variants implemented here account for the network structure, an essential aspect of community structure which is otherwise completely ignored in standard measures.