Abstract

We are inundated with data—companies such as Twitter deal with petabytes of information on a daily basis. However, some users, especially the new ones, often find it difficult to cope with the overwhelming and disorganised deluge of information. Scientists have already worked out ways to identify Twitter trending topics, as a means to index information and make sense of it. However, we know little about the impact on trending topics of various intrinsic factors associated with the Twitter ecosystem. For example, anecdotal evidence suggests that trending topics are characterised by highly polarised tweets, or that large audiences typically host the emergence of trending topics. However, no study has yet addressed these issues formally. To remedy this situation, we have launched an investigation on the nature of trending topics. Our initial observations indicate that there is a correlation between strong sentiment polarity and the emergence of trending topics—we can also confirm that the strength of the polarity drops as the trending topics fade away. Conversely, our experiments highlight that there is no correlation between the size of a Twitter audience and the rise of trending topics.

DOI

10.3233/978-1-61499-900-3-1004

Publication Date

2018-01-01

Publication Title

NEW TRENDS IN INTELLIGENT SOFTWARE METHODOLOGIES, TOOLS AND TECHNIQUES (SOMET_18)

Volume

303

Publisher

IOS Press

ISBN

978-1-61499-899-0

ISSN

1879-8314

Embargo Period

2024-11-22

Keywords

clustering, news information retrieval, sentiment analysis, TF-IDF, trending topics, Twitter

First Page

1004

Last Page

1017

Additional Files
UoP_Deposit_Agreement v1.1 20160217.pdf (124 kB)
Download

Share

COinS