Automatic tweets classification under an intelligent agents framework

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Twitter is a microblogging platform that allows users to share opinions with a restricted amount of characters. Given the social characteristic of Twitter, it is a potential source for sentiment analysis. For this reason, opinions of certain subjects such as people or brands can change in short periods of time. A traditional approach of a sentiment classifier implementation performs poorly since it depends on how it is trained. We propose a novel method for tackling this problem, with the implementation of a multi-agent system for classifying and corpus analysis mechanism for retraining the classifier. This mechanism consists of a critic agent which compares the trained corpus of the classifier agent with new collections of documents from future time steps, using primarily two methods: hypothesis test analysis and histogram differences. A Naïve-Bayes based classifier was implemented with this mechanism with multiple configurations. The results of experimental data show that the mechanism boosts its performance, when compared to a pure Naïve Bayes classifier.

Original languageEnglish
Title of host publicationSocial Computing and Social Media
Subtitle of host publicationApplications and Analytics - 9th International Conference, SCSM 2017 Held as Part of HCI International 2017, Proceedings
EditorsGabriele Meiselwitz
PublisherSpringer Verlag
Pages295-311
Number of pages17
ISBN (Print)9783319585611
DOIs
StatePublished - 2017
Externally publishedYes
Event9th International Conference on Social Computing and Social Media, SCSM 2017, held as part of the 19th International Conference on Human-Computer Interaction, HCI 2017 - Vancouver, Canada
Duration: 9 Jul 201714 Jul 2017

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10283 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference9th International Conference on Social Computing and Social Media, SCSM 2017, held as part of the 19th International Conference on Human-Computer Interaction, HCI 2017
Country/TerritoryCanada
CityVancouver
Period9/07/1714/07/17

Keywords

  • Mobile social computing and social media
  • Text classification

Fingerprint

Dive into the research topics of 'Automatic tweets classification under an intelligent agents framework'. Together they form a unique fingerprint.

Cite this