Scientific American 201907

92 Scientific American, July 2019

Vaping Hashtags Used by Humans

Vaping Hashtags Used by Bots

Behaviors and community

Products and community

Multiple-substance use

Human topic clusters

Behaviors and multiple-substance use

Behaviors and community

Products

Health, smoking cessation and multiple-substance use

Bot topic clusters

Bots tend to repeat a small number of hashtags

referring to e-cigarettes and vaping, so pairs of hashtags

are much more concentrated than in human tweets.

Topics of conversation (colors) are also more segregated.

Researchers found 137 hashtags used in 1,600 pairs.

Numerous

human

hashtags refer

to e-cigarettes

and vaping. No

pairs of hashtags

dominate, so the

usage pattern is

spread somewhat

evenly. Color clusters

represent general topics of

conversation, which often

overlap. Terms near the center are most

common to all three topic clusters. Researchers

found 238 hashtags (dots) used in 5,203 pairs (lines).

#cigars

#cigar

#cigars

#cigar

#quitsmoking#quitsmoking

#health#health

#buzz

#ecigs#ecigs

#vaping#vaping

#vapelife#vapelife

#ejuice#ejuice

#vapeporn#vapeporn

#vapenation#vapenation

#cigars#cigars

#whisky

#eliquid#eliquid

#vaping#vaping

#vapes#vapes

#weed#weed

#vapeorizer#vapeorizer

#cannabis

#vapes

#cannabis

#vapes

#vape#vape

#bigtobacco#bigtobacco

#ejuice#ejuice

#vaper#vaper

#smoke#smoke

#instavape#instavape

#vapecommunity#vapecommunity

#ecig#ecig

#vapors#vapors

#vapeclub#vapeclub #atomizer#atomizer

#vapestagram#vapestagram

#vapesociety#vapesociety

#smokers#smokers

#nowsmoking#nowsmoking

#tobacco#tobacco

#marijuana#marijuana

#scotch#scotch

#smoker#smoker

#cigarettes#cigarettes

#hookah#hookah

#addiction#addiction #blu#blu

#blu

#tobacco#tobacco

#cigaratte#cigarette

#smoke#smoke

#smoking#smoking

#photography#photography

#galenocigars#galenocigars

#life

#love

#life

#love #ecig#ecig#vapor#vapor#vape#vape

#ecigs#ecigs

#vapefam#vapefam

#vapelife#vapelife

#smok#smok

#vaporstorm

#eliquids#eliquids

#vapepen#vapepen

#vapefamily#vapefamily

#vapeshop#vapeshop

#vapecommunity#vapecommunity

#vapeporn#vapeporn#vapers#vapers

#vaporizer#vaporizer

#esmoke#esmoke

#esmoking

#online

#beast#beast

#mod#mod

#cheap#cheap

#cigpet

#starterskit#starterskit

#esmoker#esmoker

#mobile#mobile #ismog#ismog

#modbox

#marijuana#marijuana

#weed#weed

#cannabis#cannabis

#cbd#cbd

#thc#thc

#cool#cool

#bongs#bongs

#quality#quality

#ejuice#ejuice

GRAPHIC SCIENCE

Text by Mark Fischetti | Graphic by Sree Priyanka Uppu

Smoke Screen

Social media bots promote unproved

benefits of e-cigarettes

Vaping is hot. A clever analysis of

Twitter posts reveals one possible

reason: automated accounts, or bots,

may be convincing people that elec-

tronic cigarettes are beneficial.

Re searchers analyzed 2.2  million

tweets about vaping and discov-

ered that hash tags used in

tweets by humans differ from

those in tweets by bots. Bots

fo cus on new products and

on vaping as an effective way

to stop smoking tobacco even

though “there is limited scien-

tific evidence for that,” says

study leader Jon-Patrick Allem,

an assistant professor of research

at the University of Southern Califor-

nia. Hashtags written by humans em-

phasize people’s lifestyles—vaping is cool,

vapers are a community. Allem speculates the

bots are propagated by manufacturers or by orga-

nizations that promote vaper rights or that are

generally against government regulation.

Only tweets that
contained two or more
hashtags were analyzed
because hashtags occurring
together indicates they are
related. The most frequently
used hashtag terms
are noted.
Each node
( dot ) is a hashtag.
Each line connects hashtags
that occur together. The
thicker the line, the more often
the hashtag pair occurs.
Gray lines do not fall within
the primary topic clusters.
SOURCE: “E-CIGARETTE SURVEILLANCE WITH SOCIAL MEDIA DATA: SOCIAL BOTS, EMERGING TOPICS, AND TRENDS,”
BY JON-PATRICK ALLEM ET AL. IN JMIR PUBLIC HEALTH AND SURVEILLANCE, VOL. 3, NO. 4, ARTICLE E98; 2017