PreechVis: Visual profiling using multiple-word combinations

Abstract : Words in the corpus include features and information, and the visualizing of such words can improve the user's understanding of them. Those words may be consist of one-word or they may be a combination of words that together. The latter is referred to as a multiword expressions (MWEs). And if we analyze both single word and multiword with visualization, we can get more accurate results and more information than when we analyze only single word from corpus. An interactive-visualization can be useful for analyzing multiword expressions, because the following features are of interest to linguistics scholars: (1) Showing the combinations of POS pattern, (2) exploring the results according to the POS combination pattern, and (3) searching the source corpus for the verification. Therefore, we propose PreechVis, an interactive visualization tool that includes all of the requisite functions for an analysis using multiwords (http://ressources.modyco.fr/sm/PreechVisMWE/). For the present study, we used a total of 957 speeches, 164,646 sentences and 3,698,617 tokens of 43 U.S. Presidents from George Washington to Barack Obama as the corpus. PreechVis is divided into two views. In the first view, the system consists of a combination of Sunburst and RadVis. Through the Sunburst, we present the POS and its combination patterns for each gram. In RadVis, the Presidents were positioned according to their frequency value. In addition, when the President was selected, the frequency value was displayed on Sunburst to improve the user's understanding. In the second view, the user can simultaneously confirm and verify the details of the result using the Wordcloud. The two different views are synchronized each other and easy to change by the selected grams, issues, and presidents. With the experiments and case studies on the U.S. President speeches, we verified the effectiveness and usability of PreechVis.
Type de document :
Communication dans un congrès
13th International Conference on Information Visualization Theory and Applications, Jan 2018, Funchal, Portugal. SCITEPRESS - Science and Technology Publications, Proceedings of the 13th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, pp.97-107, 〈10.5220/0006615500970107〉
Liste complète des métadonnées

Littérature citée [22 références]  Voir  Masquer  Télécharger

https://hal.archives-ouvertes.fr/hal-01705493
Contributeur : Seongmin Mun <>
Soumis le : lundi 5 mars 2018 - 11:26:53
Dernière modification le : mercredi 4 juillet 2018 - 23:14:05
Document(s) archivé(s) le : mercredi 6 juin 2018 - 12:34:29

Fichier

PreechVis, Visual profiling us...
Fichiers produits par l'(les) auteur(s)

Identifiants

Collections

Citation

Seongmin Mun, Gyeongcheol Choi, Guillaume Desagulier, Kyungwon Lee. PreechVis: Visual profiling using multiple-word combinations. 13th International Conference on Information Visualization Theory and Applications, Jan 2018, Funchal, Portugal. SCITEPRESS - Science and Technology Publications, Proceedings of the 13th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, pp.97-107, 〈10.5220/0006615500970107〉. 〈hal-01705493〉

Partager

Métriques

Consultations de la notice

83

Téléchargements de fichiers

44