Sentiment analysis for brand-related social media conversations

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

Abstract

With the wide adoption of social media, and the soaring volumes of brand-related social media conversations, manual approaches to content analysis are no longer practical.
Instead, automated computational methods are now required to efficiently analyse the large volume of content data. A recent trend is to classify content according to consumers’
feelings and opinions about brands by deploying content analysis techniques for sentiment classification. We argue existing techniques used in academic research and industry practice do not fit the type of data social media provides. This study compares the lexicon-based approach to sentiment analysis with computer supervised learning approach using Facebook data. Results show the two approaches are similar in accuracy but differ substantially in their classification ensembles. To rectify the differences, this study combines the two approaches and demonstrates improved outcomes.
Original languageEnglish
Title of host publicationANZMAC 2016
Subtitle of host publicationMarketing in a post-disciplinary era : proceedings
EditorsDavid Fortin, Lucie K. Ozanne
Place of PublicationChristchurch, NZ
PublisherUniversity of Canterbury
Pages300-307
Number of pages8
ISBN (Print)9780473376604
Publication statusPublished - 2016
EventAustralia New Zealand Marketing Academy Conference - University of Canterbury, Christchurch, New Zealand
Duration: 5 Dec 20167 Dec 2016

Conference

ConferenceAustralia New Zealand Marketing Academy Conference
Abbreviated titleANZMAC
CountryNew Zealand
Period5/12/167/12/16

Keywords

  • social media
  • marketing analytics
  • sentiment analysis
  • big dat

Fingerprint

Dive into the research topics of 'Sentiment analysis for brand-related social media conversations'. Together they form a unique fingerprint.

Cite this