Impact of Social Signals in Real Time Tweet Filtering and Summarization task (International Symposium on Interdisciplinarity, Corte, 2017) - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Impact of Social Signals in Real Time Tweet Filtering and Summarization task (International Symposium on Interdisciplinarity, Corte, 2017)

Résumé

Unlike traditional data sources, the social media stream such as Tweeter is characterized by the volume, velocity, and variety of the published information, which can vary significantly in terms of quality. Filtering and summarizing the media stream for long ongoing events is a challenging task. To be effective, a trade-off between pushing too many or too few tweets need to be achieved. While the proposed approaches are based mainly on the tweet content to discard irrelevant tweets regarding the event of interest, it is unclear how effective is the use of social signals in the tweet filtering and summarization task. We investigate the impact of social signals use to evaluate the quality of tweets. Two kinds of social signals are considered: the first ones are related to the author and the second ones are tweet specific features. Experiments were carried out on two datasets namely TREC MB Real Time Filtering 2015 and TREC MB Real Time Summarization 2016. The experiments that we conducted show the interest of integrating social signal and the use of machine learning algorithm to improve the quality of real-time tweet filtering approaches.

Mots clés

Fichier non déposé

Dates et versions

hal-03131898 , version 1 (04-02-2021)

Identifiants

  • HAL Id : hal-03131898 , version 1

Citer

Abdelhamid Chellal, Bernard Dousset. Impact of Social Signals in Real Time Tweet Filtering and Summarization task (International Symposium on Interdisciplinarity, Corte, 2017). International Symposium on Interdisciplinarity 2017, Fédération de Recherche Environnement et Société (FRES) de l’Université de Corse; CNRS; INRA; INSERM, Jul 2017, Corte, France. ⟨hal-03131898⟩
33 Consultations
0 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More