Nos tutelles

CNRS UNS

Nos partenaires

ILF MSHS-SE

Rechercher




Accueil > Pages Perso > Laurent Vanni

Laurent Vanni

IE -  CNRS

I - Compétences Techniques

Langages : Python, C, PHP, HTML, CSS
Technologies : HTML5, CSS3, JQuery, jSon, No-SQL DB
Software : Linux, OSX, Git
Langues : Français, Anglais

II - Experience

Jan 2013 - Aujourd’hui : Ingénieur d’Etude, UMR 7320 Bases Corpus Langage, CNRS France.
Ingénieur en développement et déploiement d’applications

Oct 2009 - Dec 2012 : Ingénieur Expert, INRIA, Lognet Team, Sophia-Antipolis, France.
Ingénieur Leader du projet Européen myMed (INTERREG IV - Alcotra) : Création d’un Réseaux Social Transfrontalier.
Architecte logicielle, responsable technique, analyses des besoins, analyses fonctionnelles
Clients : CG06, CARF Menton, ville de Cuneo
Technologies : PHP5, JQuery, jSon, JEE, No-SQL DB Cassandra

Oct 2008 - Sep 2009 : Stagiaire, INRIA, OASIS Team, Sophia Antipolis, France.
Développement de pluggins JAVA pour l’IDE Eclipse RCP.
Optimiser le debugger d’Eclipse et ajouter des fonctionnalités à l’IDE pour améliorer le debuggage des appli- cations distribuées liées au framework ProActive.
Technologies : Applications Distribuées, multi-Threading, Java, Eclipse RCP

Mai 2008 - Oct 2008 : Stagiaire, INRIA, OASIS Team, Sophia Antipolis, France.
Fournir une suite d’outils en JAVA pour faciliter le contrôle de l’exécution d’applications réparties utilisant ProActive (notamment un mode Pas à Pas).
Technologies : Applications Distribuées, multi-Threading, Java, Eclipse RCP

III - Education

2009 : Master 2 IFI , Université de Nice Sophia-Antipolis, France. Système, Sécurité et Réseaux (AB).

2007 : Licence , Université de Nice Sophia-Antipolis, France. Software Engineering (AB).

Dernières publicationsHAL

pour l'idHal "lvanni" :

titre
Machine Learning under the light of Phraseology expertise: use case of presidential speeches, De Gaulle -Hollande (1958-2016)
auteur
Mélanie Ducoffe, Damon Mayaffre, Frédéric Precioso, Frédéric Lavigne, Laurent Vanni, A Tre-Hardy
article
Damon Mayaffre; Céline Poudat; Laurent Vanni; Véronique Magri; Peter Follette. JADT 2016 - Statistical Analysis of Textual Data, Jun 2016, Nice, France. Presses de FacImprimeur, JADT - Statistical Analysis of Textual Data, Volume 1, pp.157-168, 2016, JADT 2016 - Statistical Analysis of Textual Data. 〈https://jadt2016.sciencesconf.org/〉
annee_publi
2016
resume
Author identification and text genesis have always been a hot topic for the statistical analysis of textual data community. Recent advances in machine learning have seen the emergence of machines competing state-of-the-art computational linguistic methods on specific natural language processing tasks (part-of-speech tagging, chunking and parsing, etc). In particular, Deep Linguistic Architectures are based on the knowledge of language speci-ficities such as grammar or semantic structure. These models are considered as the most competitive thanks to their assumed ability to capture syntax. However if those methods have proven their efficiency, their underlying mechanisms, both from a theoretical and an empirical analysis point of view, remains hard both to explicit and to maintain stable, which restricts their area of applications. Our work is enlightening mechanisms involved in deep architectures when applied to Natural Language Processing (NLP) tasks. The Query-By-Dropout-Committee (QBDC) algorithm is an active learning technique we have designed for deep architectures: it selects iteratively the most relevant samples to be added to the training set so that the model is improved the most when built from the new training set. However in this article, we do not go into details of the QBDC algorithm-as it has already been studied in the original QBDC article-but we rather confront the relevance of the sentences chosen by our active strategy to state of the art phraseology techniques. We have thus conducted experiments on the presidential discourses from presidents C. De Gaulle, N. Sarkozy and F. Hollande in order to exhibit the interest of our active deep learning method in terms of discourse author identification and to analyze the extracted linguistic patterns by our artificial approach compared to standard phraseology techniques.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.archives-ouvertes.fr/hal-01343209/file/JADT2016_Ducoffe_et_al.pdf BibTex
titre
How to explore conflicts in French Wikipedia talk pages?
auteur
Céline Poudat, Laurent Vanni, Natalia Grabar
article
Damon Mayaffre; Céline Poudat; Laurent Vanni; Véronique Magri; Peter Follette. Statistics Analysis of Textual Data, Jun 2016, Nice, France. JADT2016 International Conference on Statistical Analysis of Textual Data, 2, pp.645-656, 2016, JADT 2016 - Statistical Analysis of Textual Data
annee_publi
2016
resume
With the exponential development of the Internet, new discourse genres and situations have expanded. These new web genres, which are still little described, are complex objects challenging our methodologies and our analysis tools: the encyclopedic project Wikipedia is one of these new objects which are part of Computer-mediated communication (CMC). The present article concentrates on the exploration of conflicts in Wikipedia talk pages, using Hyperbase Web. Wikipedia data and CMC corpora have been little studied by French linguistics so far, and are still challenging text statistics, notably because of the complexity of such data (multiple annotations, consistent metadata, references between postings and user networks). Based on the Wikiconflits corpus, which is already available and freely usable by researchers, we will propose some methodological avenues to explore Wikipedia data and CMC corpora.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.archives-ouvertes.fr/hal-01359416/file/78404.pdf BibTex
titre
Cooccurrences spécifiques et représentations graphiques, le nouveau " Thème " d'Hyperbase
auteur
Laurent Vanni, Adiel Mittmann
article
Damon Mayaffre; Céline Poudat; Laurent Vanni; Véronique Magri; Peter Follette. JADT 2016 - Statistical Analysis of Textual Data, Jun 2016, Nice, France. Presses de FacImprimeur, JADT 2016 - Statistical Analysis of Textual Data, 1, pp.295-305, 2016, Statistics Analysis of Textual Data
annee_publi
2016
resume
Finding word cooccurrences and calculating the specificity scores is one of the most popular statistical methods in the analysis of textual data. Within Hyperbase, there is a " theme " feature for this purpose, which is capable of locating words that are used more commonly near a given word form, grammatical structure or lemma. The graphical representation of such an analysis is often challenging ; more than a list of the cooccurring words, it should be able to indicate the order, the score and the relations between pairs of words. Now that Hyperbase has a Web version, this article proposes a new approach for the " theme " feature : the calculation of cooccurrences has been extended to include the second level. The accompanying graphical representation is betting on new visual features in order to simplify reading the results and render the underlying calculation more explicit. Résumé Le calcul des cooccurrents spécifiques d'un mot est une des méthodes statistiques les plus populaires de l'ADT. Le logiciel Hyperbase a introduit cette notion avec la fonction " Th eme " qui rend compte du lexique surutilisé autour d'une forme, d'une structure grammaticale ou d'un lemme. La représentation graphique d'une telle analyse est souvent loin d'ˆ etré evidente. Plus que la simple liste des cooccurrents spécifiques , elle se doit de nous indiquer aussi l'ordre, l'´ ecart ainsi que les relations entre chaque paire de mots. Avec l' arrivée de la version Web d'Hyperbase, nous proposons aujourd'hui une nouvelle approche de cette fonction. Le calcul se voit pour l'occasion approfondi, lui permettant d'identifier la cooccurrence dedeuxì eme niveau. La représentation graphique, quant a elle, fait le pari de simplifier la lecture du résultat de cette analyse tout en explicitant plus précisément les calculs sous-jacents.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.archives-ouvertes.fr/hal-01359413/file/85261.pdf BibTex
titre
JADT 2016 - Statistical Analysis of Textual Data
auteur
Damon Mayaffre, Céline Poudat, Laurent Vanni, Véronique Magri, Peter Follette
article
Damon Mayaffre; Céline Poudat; Laurent Vanni; Véronique Magri; Peter Follette. JADT 2016 - Statistical Analysis of Textual Data, Jun 2016, Nice, France. 2, Presses de FacImprimeur, 862 p., 2016, 978-2-7466-9067-7
annee_publi
2016
resume
Proceedings of 13 th International Conference on Statistical Analysis of Textual Data 7-10 June 2016 - Nice
typdoc
Direction d'ouvrage, Proceedings, Dossier
Accès au bibtex
BibTex
titre
Arbre et co-occurrences Nouvel outil logométrique sur le net. Application au discours de François Hollande
auteur
Laurent Vanni, Xuan Luong, Damon Mayaffre
article
Emilie Née; Jean-Michel Daube; Mathieu Valette; Serge Fleury. JADT 2014, Jun 2014, Paris, France. p. 639-649
annee_publi
2014
resume
A partir de 9 matrices mots x mots ou matrices co-occurrentielles (une par man-dat présidentiel depuis 1958), nous produisons une matrice de dissimilarité consignant les distances entre les présidents de la Vème République. On donne une représentation ar-borée de cette matrice et on améliore ici les performances de la représentation grâce à une nouvelle approche topologique. On présente alors l'outil logiciel qui permet de tracer le graphe et au terme du parcours méthodologique, on produit une première description socio-linguistique du discours de François Hollande qui reste selon les observateurs poli-tiques encore difficile à définir dans l'histoire présidentielle française. Abstract From 9 matrices of co-occurrences (one for each french president since 1958), we produce a dissimilarities matrix of distances between presidents of the fifth republic. We give a tree analysis of this matrix and we improve the performances due a new topo-logical approach. We present a new software that provide a graphical visualization with this tree analysis. Finally we suggest a first socio-linguistic description of the François Hollande discourse which stay, according to political reporters, still difficult to define in the presidential history.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.archives-ouvertes.fr/hal-01196592/file/jadt2014_latex.pdf BibTex
titre
GOOFRE version 2
auteur
Etienne Brunet, Laurent Vanni
article
Emilie Née; Jean-Michel Daube; Mathieu Valette; Serge Fleury. JADT 2014, Jun 2014, Paris, France. p. 106-119
annee_publi
2014
resume
The amount of data contained within Google Books has doubled over the last two years and now exceeds 500 billion words. A new treatment of the data has included a re-examination of scanned images, offering a more accurate recognition of the text. In addition, for the first time, included texts have been subjected to deambigation and lemmatisation. Finally, the website Culturomics has made tools available that facilitate its accessibility. It seemed interesting, therefore, to develop a new expertise and to create a new database, complete with all the necessary statistical tools, available online or locally, for exploiting such large corpora.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.archives-ouvertes.fr/hal-01196595/file/jadt2014-paper-62.pdf BibTex
titre
Synapse: A Scalable Protocol for Interconnecting Heterogeneous Overlay Networks
auteur
Luigi Liquori, Cédric Tedeschi, Laurent Vanni, Francesco Bongiovanni, Vincenzo Ciancaglini, Bojan Marinkovic
article
Mark Crovella and Laura Marie Feeney and Dan Rubenstein and S. V. Raghavan. NETWORKING 2010 9th International IFIP TC 6 Networking Conference, Chennai, India, May 11-15, 2010. Proceedings, May 2010, Chennai, India. Springer Verlag, 6091, pp.67-82, 2010, Lecture Notes in Computer Science. 〈10.1007/978-3-642-12963-6_6〉
annee_publi
2010
resume
This paper presents Synapse, a scalable protocol for information retrieval over the inter-connection of heterogeneous overlay networks. Applications on top of Synapse see those intra-overlay networks as a unique inter-overlay network. Scalability in Synapse is achieved via co-located nodes, i.e. nodes that are part of multiple overlay networks at the same time. Co-located nodes, playing the role of neural synapses and connected to several overlay networks, give a larger search area and provide alternative routing. Synapse can either work with “open” overlays adapting their protocol to synapse interconnection requirements, or with “closed” overlays that will not accept any change to their protocol. Results from simulation and experiments show that Synapse is scalable, with a communication and state overhead scaling similarly as the networks interconnected. thanks to alternate routing paths, Synapse also gives a practical solution to network partitions. We precisely capture the behavior of traditional metrics of overlay networks within Synapse and present results from simulations as well as some actual experiments of a client prototype on the Grid’5000 platform. The prototype developed implements the Synapse protocol in the particular case of the inter-connection of many Chord overlay networks.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.inria.fr/hal-00909544/file/synapse.pdf BibTex
titre
Synapse: a Scalable Protocol for Interconnecting Heterogeneous Overlay Networks
auteur
Luigi Liquori, Cédric Tedeschi, Laurent Vanni, Vincenzo Ciancaglini, Francesco Bongiovanni, Bojan Marinkovic
article
[Research Report] RR-7255, INRIA Sophia Antipolis - Méditerranée; INRIA Rennes - Bretagne Atlantique; INRIA. 2010, pp.16
annee_publi
2010
resume
This paper presents Synapse, a scalable protocol for information retrieval over the inter-connection of heterogeneous overlay networks. Applications on top of Synapse see this collection of intra-overlay networks as a unique inter-overlay network. Scalability in Synapse is achieved via co-located nodes, i.e., nodes that are part of multiple overlay networks at the same time. Co-located nodes, playing the role of neural synapses and connected to several overlay networks, provide a larger search area and provide alternative routing. Synapse can either work with "open'' overlays adapting their protocol to synapse interconnection requirements, or with "closed'' overlays that will not accept any change to their protocol. Results from simulation and experiments show that Synapse is scalable, with a communication and state overhead scaling similarly as the networks interconnected. thanks to alternate routing paths, Synapse also gives a practical solution to network partitions. We precisely capture the behavior of traditional metrics of overlay networks within Synapse and present results from simulations as well as some actual experiments of a client prototype on the Grid'5000 platform. The prototype developed implements the Synapse protocol in the particulare case of the inter-connection of many Chord overlay networks.
typdoc
Rapport
Accès au texte intégral et bibtex
https://hal.inria.fr/inria-00474529/file/RR-7255.pdf BibTex
titre
CarPal: interconnecting overlay networks for a community-driven shared mobility
auteur
Vincenzo Ciancaglini, Luigi Liquori, Laurent Vanni
article
Martin Wirsing and Martin Hofmann and Axel Rauschmayer. Trustworthly Global Computing 5th International Symposium, TGC 2010, Munich, Germany, February 24-26, 2010, Revised Selected Papers, Feb 2010, München, Germany. Springer Verlag, 6084, pp.301-317, 2010, Lecture Notes in Computer Science. 〈10.1007/978-3-642-15640-3_20〉
annee_publi
2010
resume
Car sharing and carpooling have proven to be an effective solution to reduce the amount of running vehicles by increasing the number of passengers per car amongst medium/big communities like schools or enterprises. However, the success of such practice relies on the community ability to effectively share and retrieve information about travelers and itineraries. Structured overlay networks such as Chord have emerged recently as a flexible solution to handle large amount of data without the use of high-end servers, in a decentralized manner. In this paper we present CarPal, a proof-of-concept for a mobility sharing application that leverages a Distributed Hash Table to allow a community of people to spontaneously share trip information without the costs of a centralized structure. The peer-to-peer architecture allows moreover the deployment on portable devices and opens new scenarios where trips and sharing requests can be updated in real time. Using an original protocol already developed that allows to interconnect different overlays/communities, the success rate (number of shared rides) can be boosted up thus increasing the effectiveness of our solution. Simulations results are shown to give a possible estimate of such effectiveness.
typdoc
Communication dans un congrès
Accès au texte intégral et bibtex
https://hal.inria.fr/hal-00909531/file/CarPal.pdf BibTex
  • + de résultats dans la Collection HAL du laboratoire BCL
  • Voir l'ensemble des résultats sur la plateforme HAL