Layout 1


ISDS Annual Conference Proceedings 2012. This is an Open Access article distributed under the terms of the Creative Commons Attribution-
Noncommercial 3.0 Unported License (http://creativecommons.org/licenses/by-nc/3.0/), permitting all non-commercial use, distribution, and
reproduction in any medium, provided the original work is properly cited.

ISDS 2012 Conference Abstracts

Content Analysis of Tobacco-related Twitter Posts
Mark Myslín*, Shu-Hong Zhu and Michael Conway

UC San Diego, San Diego, CA, USA

Objective
We present results of a content analysis of tobacco-related Twitter

posts (tweets), focusing on tweets referencing e-cigarettes and
hookah.

Introduction
Vast amounts of free, real-time, localizable Twitter data offer new

possibilities for public health workers to identify trends and attitudes
that more traditional surveillance methods may not capture, particu-
larly in emerging areas of public health concern where reliable sta-
tistical evidence is not readily accessible. Existing applications
include tracking public informedness during disease outbreaks [1].

Twitter-based surveillance is particularly suited to new challenges
in tobacco control. Hookah and e-cigarettes have surged in popular-
ity, yet regulation and public information remain sparse, despite con-
troversial health effects [2,3]. Ubiquitous online marketing of these
products and their popularity among new and younger users make
Twitter a key resource for tobacco surveillance.

Methods
We collected 7,300 tobacco-related Twitter posts at 15-day inter-

vals from December 2011 to July 2012, using ten general keywords
such as cig* and hookah. Each tweet was manually classified using
a tri-axial scheme, capturing genre (firsthand experience, joke, news,
…), theme (underage usage, health, social image, …), and sentiment
(positive, negative, neutral). Machine-learning classifiers were
trained to detect tobacco-related vs. irrelevant tweets as well as each
of the above categories, using Naïve Bayes, k-Nearest Neighbors,
and Support Vector Machine algorithms. Finally, phi correlation co-
efficients were computed between each of the categories to discover
emergent patterns.

Results
The most prevalent genre of tweets was personal experience, fol-

lowed by categories such as opinion, marketing, and news. The most
common themes were hookah, cessation, and social image, and sen-
timent toward tobacco was more positive (26%) than negative (20%).
The most highly correlated categories were social image–underage,
marketing–e-cigs, and personal experience–positive sentiment. E-cig-
arettes were also correlated with positive sentiment and new users
(even excluding marketing posts), while hookah was highly corre-
lated with positive sentiment, pleasure, and social relationships. Fur-
ther, tweets matching the term “hookah” reflected the most positive
sentiment, and “tobacco” the most negative (Figure 1). Finally, neg-
ative sentiment correlated most highly with social image, disgust, and
non-experiential categories such as opinion and information.

The best machine classification performance for tobacco vs. non-
tobacco tweets was achieved by an SVM classifier with 82% accu-
racy (baseline 57%). Individual categories showed similar
improvements over baseline.

Conclusions
Several novel findings speak to the unique insights of Twitter sur-

veillance. Sentiment toward tobacco among Twitter users is more
positive than negative, affirming Twitter’s value in understanding
positive sentiment. Negative sentiment is equally useful: for example,

observed high correlations between negative sentiment and social
image, but not health, may usefully inform outreach strategies. Twit-
ter surveillance further reveals opportunities for education: positive
sentiment toward the term “hookah” but negative sentiment toward
“tobacco” suggests a disconnect in users’ perceptions of hookah’s
health effects. Finally, machine classification of tobacco-related posts
shows a promising edge over strictly keyword-based approaches, al-
lowing for automated tobacco surveillance applications.

Sentiment in “hookah” tweets is disproportionately more positive than in
“cig” and especially “tobacco” tweets.

Keywords
social media; surveillance; Twitter; tobacco; NLP

References

[1] Chew, C. & Eysenbach, G. Pandemics in the age of twitter: Content
analysis of tweets during the 2009 H1N1 outbreak. PLoS One. 2010;
5(11):e14118. 

[2] Ayers, J. W., Ribisl, K. M., & Brownstein, J. S. Tracking the rise in
popularity of electronic nicotine delivery systems (Electronic ciga-
rettes) using search query surveillance. Am J Prev Med. 2011;
40(4):448–453. 

[3] Grekin, E. R. & Ayna, D. Waterpipe smoking among college students
in the United States: A review of the literature. J Am Coll Health.
2012; 60(3):244–249.

*Mark Myslín
E-mail: mmyslin@gmail.com

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * 5(1):e66, 2013