Annonce en lien avec l’Action/le Réseau : Doctorants
Laboratoire/Entreprise : IECL
Durée : 3 years
Contact : marianne.clausel@univ-lorraine.fr
Date limite de publication : 2019-05-29
Contexte :
Context : Social networks and medias in general create a huge quantity of information which may differ according to the location (countries, areas, cities…..)
and the time periods. A natural question is to identify which main topics are
persistent in a corpus of documents as tweets, websites or scientific papers. The
aim of the project is to take into account the specifities of data as similarities
between different regions or countries as well as the time stamp of the document…This question has been already addressed in several papers (see for e.g. [1] and several models have been proposed to summarize the temporal evolution
(see for e.g. [2]).
– [1] S. Asur, B. A. Huberman, G. Szabo, C. Wang. Trends in social media: persistence and decay. In ICWSM. (2011).
– [2] Y. Wang, E. Agichtein, M. Benzi. TM-LDA: efficient online modeling of latent
topic transitions in social media. Proc. of the 18th ACM SIGKDD. ACM (2012).
Sujet :
Challenges : We aim at complementing these works studying spatio-temporal
persistence in textual data. Using dynamic topic modeling [3], we can modeled
in real-time the content evolution of a corpus. Our goal will be to identify which
topics are persistent in a corpus, taking into account both spatial and temporal
information. The part simulation and inference will be designed using Monte
Carlo methods [6,7] whereas persistence will be measured using multivariate
long range dependence [4].
– [3] D. Blei, J. D. Lafferty. Dynamic topic models. Proceedings of the 23rd international conference on Machine learning. ACM, (2006).
– [4] S. Kechagias, V. Pipiras. Definitions and representations of multivariate longrange dependent time series. JTSA 36.1 1-25 (2015).
– [5] M. Li, X. Wang, K. Gao, S. Zhang. A survey on information diffusion in
online social networks: Models and methods. Information 8, no. 4: 118 (2017).
– [6] G. Winkler, Image analysis, random fields and MCMC methods, Springer (2003)
– [7] R. S. Stoica, A. Philippe, P. Gregori, J. Mateu. ABC Shadow algorithm: a
tool for statistical analysis of spatial patterns. Stat. Comp., 27(5) : 1225-1238, (2017
Profil du candidat :
We are seeking a candidate having strong skills in Probability and Statistics, as well as programming in Python
Formation et compétences requises :
Master M2 in Probability and Statistics or in Computer Sciences
Adresse d’emploi :
Instititu Elie Cartan de Lorraine
Nancy