Abstract: The construction of the ETL (Extract-Transform-Load) process is one of the biggest tasks of building a warehouse. ETL processes area has little research, because of its difficulty and lack of formal model for representing ETL activities that map the incoming data from different sources to be in a suitable format for loading into the warehouse. A main problem in data warehousing of multidimensional text databases is to deal with the content in its text cells. In this paper, we propose a model for textual data warehouse ETL processes called ETL-Text. It combines classical data warehousing tasks, information retrieval (IR) techniques, and information processing in particular the language modeling. Our approach is based on Wikipedia as external knowledge source to extract the semantics of the textual documents. To validate our approach, we develop a prototype composed of several processing modules that illustrate the different ETL-Text processes. Also, we use the 20 Newsgroups corpus to perform our experimentation.

Abstract: The objective of the thesis is to explore how complex data can be treated using unsupervised machine learning techniques, in which additional information is injected to guide the exploratory process. Starting from specific problems, our contributions take into account the different dimensions of the complex data: their nature (image, text), the additional information attached to the data (labels, structure, concept ontologies) and the temporal dimension. A special attention is given to data representation and how additional information can be leveraged to improve this representation.

---------

Abstract: Online social networks play a major role in the spread of information at very large scale and it becomes essential to provide means to analyze this phenomenon. Analyzing information diffusion proves to be a challenging task since the raw data produced by users of these networks are a flood of ideas, recommendations, opinions, etc. The aim of this PhD work is to help in the understanding of this phenomenon. So far, our contributions are the following: (i) a survey of developments in the field; (ii) T-BaSIC, a graph-based model for information diffusion prediction; (iii) SONDY, an open source platform that helps understanding social network users' interests and activity by providing emerging topics and events detection as well as network analysis functionalities.

---------

Abstract: In this paper we examine the interest around a number of television series broadcast on a weekly basis. We show that through analysis of initial conversation between fans or users of dedicated web forums we can provide a description of the greatest period of interest (or peak). We then focus our attention on this peak with an ultimate goal of characterising episodes as a function of the actions and qualities of the people that take part in early conversation about this episode. We find that early interaction trends have strong similarities with the overall conversation patterns, and contain the majority of information provided by influential members of the community. This observation has important implications for the rapid generation of meta-data which may be used to during later broadcast and re-runs for description and valuation of episodes.

Abstract: Le magazine “L’Etudiant” a publié en septembre 2012 un classement des “villes où il fait bon étudier”, présentant un palmarès multi-critère des 41 villes françaises de plus de 8000 étudiants. Nous proposons ici une étude visant à analyser ce palmarès à travers des méthodes d’agrégation diversifiées, afin de faire ressortir les invariants d’un tel classement de villes, et au contraire des effets directement liés à l’utilisation de telle ou telle méthode.

---------

DOI: 10.1016/j.cor.2013.04.003
Abstract: We study in this paper the generation of the Choquet optimal solutions of biobjective combinatorial optimization problems. Choquet optimal solutions are solutions that optimize a Choquet integral. The Choquet integral is used as an aggregation function, presenting dierent parameters, and allowing to take into account the interactions between the objectives. We develop a new property that characterizes the Choquet optimal solutions. From this property, a general method to easily generate these solutions in the case of two objectives is dened. We apply the method to two classical biobjective optimization combinatorial optimization problems: the biobjective knapsack problem and the biobjective minimum spanning tree problem. We show that Choquet optimal solutions that are not weighted sum optimal solutions represent only a small proportion of the Choquet optimal solutions and are located in a specic area of the objective space, but are much harder to compute than weighted sum optimal solutions.

---------

Abstract: De nos jours, les entrepôts de données et les outils d’analyse OLAP sont très utilisés dans les entreprises qui ont besoin de systèmes décisionnels qui s’adaptent à toutes les situations particulières du monde réel, pour éviter les erreurs d’analyse (plus connus dans la littérature sous le nom de problèmes d’additivité ou summarizability issues en Anglais). Dans cet article, nous présentons un état de l’art des travaux sur les problèmes d’additivité dus à la présence des hiérarchies complexes dans un modèle multidimensionnel et discutons des travaux restant à mener dans ce domaine.