3. !
A
tool
to
help
companies
determine
what
is
being
said
about
their
brand,
product
or
service
on
social
media
Social
Media
Analysis
Metrics
Convert
unstructured
information
from
user
generated
content
into
meaningful
consumer
understanding
Number
of
comments
or
posts
Sentiment
!
Volume
Overall
tone,
polarity
to
brand,
product
Topics
Features
and
words
associated
with
brand
Influence
Ability
of
individual
to
create
buzz
3
4. ! Thai
Natural
Language
Processing
! Word
segmentation
! Word
normalization
! Text
Mining
! Feature
word
extraction
! Topic
filtering/classification
! Sentiment
Analysis
4
13. Social
Media
for
Emergency
Response
http://mashable.com/2011/02/11/social-‐media-‐in-‐emergencies/
13
14. ! The
2011
flood
(July
–
December
2011)
has
been
described
as
the
country’s
worst
disaster
in
recent
history
! The
World
Bank
estimated
total
economic
damage
to
be
approximately
$45
billion
! The
National
Social
and
Economic
Development
Board
reduced
Thailand’s
economic
growth
forecast
from
3.5-‐4%
to
1.5%
14
17. ! Retrieved
175,551
Tweets
(from
23/10/2011
to
17/12/2011)
using
the
keyword
‘#thaiflood’
! After
removing
“Retweets”
and
duplicates,
our
data
set
contained
64,582
unique
Tweets
17
18. ! Situational
announcements
and
alerts
! up-‐to-‐date
situational
and
location-‐based
information
related
to
the
flood
such
as
water
levels,
traffic
conditions
and
road
conditions
in
certain
areas.
In
addition,
emergency
warnings
from
authorities
advising
citizens
to
evacuate
areas,
seek
shelter
or
take
other
protective
measures
are
also
included.
! Support
announcements
! support
announcements
such
as
free
parking
availability,
free
emergency
survival
kits
distribution
and
free
consulting
services
for
home
repair.
18
19. ! Requests
for
assistance
! messages
requesting
any
types
of
assistance;
such
as
food,
water,
medical
supplies,
volunteers
or
transportation.
! Requests
for
information
! general
inquiries
related
to
the
flood
and
flood
relief
such
as
inquiries
for
telephone
numbers
of
relevant
authorities,
regarding
the
current
situation
in
specific
locations
and
about
flood
damage
compensation.
! Other
! all
other
messages,
such
as
general
comments;
complaints
and
expressions
of
opinions.
19
24. ! Ranking
the
users
based
on
the
number
of
Retweets
can
help
identify
the
perceived
credibility
of
the
influencers
! With
this
list
of
influential
Twitter
users,
citizens
can
choose
which
sources
of
information
they
would
follow
during
the
natural
disaster;
in
order
to
obtain
the
most
relevant,
up-‐to-‐date,
and
credible
information
24
25. ! Fast
dissemination
of
information
to
large
groups
! Interactive
! Access
through
mobile
devices
! Growing
natural
proclivity
25
26. ! Aggregation
(lots
of
tiny
pieces!)
! Anomaly
detection
(what
isn’t
just
noise?)
! Accurate
information
(vs.
outdated,
inaccurate,
and
false
information)
! Malicious
use
of
social
media
during
disasters
26
27. ! Language
Analysis
and
Text
Mining
! Blog
Analysis
! Comparison
detection
! Sarcasm
detection
! Rumor
detection
27