GVU Brown Bag Seminar, Georgia Tech 12/6
emotion (Golder & Macy,
2011; De Choudhury et al,
2012)
information
diffusion (Kempe,
Kleinberg, 2003; Sun, Marlow,
2009; De Choudhury et al.
2010)
.
social roles (Agarwal et
al, 2008; Naaman et al, 2010)
network structure (Leskovec et al 2010)
crisis management (Starbird et al 2010)
social influence (Bakshy et al, 2011)
social ties (Gilbert & Karahalios,
2009)
.
.
.
groups and
communities (Kumar et al, 2006; Burke
et al, 2009; De
Choudhury et al. 2010)
citizen journalism (Diakopoulos & Shamma, 2010;
De Choudhury et al 2012)
urban informatics (Crandall et al, 2009; Cranshaw
et al, 2012)
what I have been doing
human behavior
online o how people spread
information
o how people form groups
o how shared social
artifacts evolve
o how people express
emotion
o mining frameworks
for deciphering
human behavior
o predictive models
o analytic insights
from big data
o improve the
experience of the
user
o understand the
impact of these
technologies on
their behavior
understanding individual behavior
It is getting harder to hide pain
examine patterns of activity and
emotional correlates for childbirth
and postnatal course leveraging
activity in social media
(De Choudhury, Counts, Horvitz, CSCW 2013)
low-cost, privacy-preserving
mechanisms to identify new mothers’
behavior can improve social support
and encourage postpartum wellness
(De Choudhury, Counts, Horvitz, CSCW 2013)
(1) birth, weigh*, pounds/lbs, inches, length/long,
baby/son/daughter/boy/girl
(2) announc*, birth of, son/daughter/brother/sister
(3) announc*, arrival of, son/daughter/brother/sister
(4) are the parents of, son/daughter/boy/girl/baby
(5) welcome* home by, brother/sister/sibling*
(6) is the proud big brother/sister
(7) after, labor, born
(8) it’s a boy/girl, born
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
prenatal-postnatal comparison
blue line represents approximate time of childbirth.
The beige line represents mothers and the green line
represents the background cohort
(De Choudhury, Counts, Horvitz, CSCW 2013)
Mothers after childbirth (MA)
[high NA] Ugh, my daughter hates her bassinet. I hate disappointing her. What a miserable day.
[low activation] My baby is only catnapping during the day. That’s so sad and depressing. I feel
helpless
[low dominance] Anxiety/panic attacks need to eff off!!!!!!!!!!!!!! I’m trying to lead a somewhat
normal life with my baby!!!! #frustrated #miserable
[high 1st person pronoun use] No lie I fuckin miss all socializing..... my daughter keeps me
occupied and exhausted. I have all my moments of the day
Mothers before childbirth (MP)
Derek & I sat on our screened in back porch listening to the thunderstorm & rain! So peaceful!
Just to think in 35 hours we’ll be parents!
Pregnant for the first time and I’m afraid I won’t be able to stand the labor pain. Husband trying
to reassure me, but he seems scared too. Thoughts????
I’m completely thrilled at the prospect of becoming a mother but the weight gain is bothering
me :(:(. Do I just need to get over myself'? Am I the only one :S
Days are getting busy!!! Need to start packing for the hospital, in case the baby is coming early!
(De Choudhury, Counts, Horvitz, CSCW 2013)
Microsoft Research
(De Choudhury, Counts, Horvitz, CSCW 2013)
measure small effect medium effect large effect
activity 38 29 20
emotion 17 4 12
style 43 3 18
effect sizes (based on Cohen’s d) over the three types of measures.
Numbers indicate the number of new mothers showing changes following
childbirth of each effect size.
(De Choudhury, Counts, Horvitz, CSCW 2013)
Mothers w/ small effects
I know some drs say it’s ok to be on meds while breastfeeding but it kind of freaks me out cause it
isn't proven longterm for baby's health.
Days are passing by as I watch my son grow! Can’t wait for more and get together with the daddy!!
Wish he was here
Just adjusting to having a new baby, new job and we just moved town. Need to calm down.
Tips/suggestions on parenting, mothers??
I'm taking expressed breastmilk from the fridge on outings in the diaper bag and keeping it cool with
an ice pack. Someone tried it?
Mothers w/ large effects
Starting to feel lost. I’m missing my love, my baby. Feel angry n disappointed in myself. Idk what to
think or do....
My first time being alone with my baby and I cant stop crying. What is wrong with me? Am I
depressed? Im just over here balling my eyes out
My DS doesnt sleep more than 3 hrs at a time and cries often and is so difficult to calm down. Cant
remember when was the last time I slept
Feel like having a breakdown! ...like the WORST mother... feel so terribly that this poor child is stuck
with this horrible monster mother..
(De Choudhury, Counts, Horvitz, CSCW 2013)
size of vocabulary
(#unigrams)
percentage of unigrams
w/ p < .01
mothers w/ large effects 17,117 27.65%
mothers w/ small effects 33,785 19.74%
background cohort 47,214 3.86%
percent of all unigrams in the language vocabulary used that changed significantly in usage frequency
after childbirth, i.e. with p < .01 based on paired sample t-tests
background cohort mothers w/ small effects mothers w/ large effects
now (↓), shit (↑), back (↑), that (↑), day
(↓), life (↑), time (↓), them (↑), me (↑),
you (↑), fuck (↑), today (↓), sleep (↑),
tonight (↓), love (↓), good (↓), here(↓),
her (↓), morning (↑), tomorrow (↑), go
(↑), know (↑), him (↓), people (↓)
#past (↑), duh (↑), people (↓),
photo (↑), post (↑), decision (↓),
reunite (↓), women (↑), story (↑),
time (↑), asap (↓), do (↑), life (↓),
wait (↑), fired (↑), days (↑), happy
(↓)
haha (↓), blessed (↑), lol (↓),
#lifecangetbetter (↑), awesome
(↓), monthly (↑), fantastic (↓),
cuddle (↑), home (↑), love (↓), sick
(↑), aww (↑), scary (↑)
top unigrams showing the most change (in usage frequency) in the postnatal period, compared to the
prenatal phase
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
predict
(De Choudhury, Counts, Horvitz, CSCW 2013)
(Corinna & Vapnik, 1995)
(De Choudhury, Counts, Horvitz, CSCW 2013)
Measures C1 C0 Measures C1 C0
volume -0.878 0.838 verbs -0.911 0.951
Replies -0.983 1.417 aux-verbs -0.889 0.914
RT -0.847 2.188 adverbs 3.025 -0.678
Links -0.791 2.347 preposition -0.899 1.020
Questions 2.110 -0.749 conjunction -0.890 1.042
PA -0.798 0.428 negate 2.867 -0.690
NA 1.726 -0.593 quantifier -0.896 0.950
activation -0.743 0.525 swear 3.258 -0.751
dominance -0.717 0.543 tentative 3.710 -0.700
Followers -0.941 1.451 certain -0.997 1.386
Followees -0.928 1.013 inhibition 3.425 -0.722
functional
words 4.574 -0.865 inclusive -0.914 1.048
1st pp. 1.924 -0.426 exclusive -0.822 2.304
2nd pp. -0.901 0.872 assent -0.904 1.073
3rd pp. -0.894 0.990 non-fluency 3.028 -0.724
indefinite
pronoun -0.810 2.082 filler 2.929 -0.786
article 3.940 -0.701
Median changes in different measures for the two classes of new mothers with
C1 corresponding to extreme-changing mothers and C0 corresponding to
mothers showing smaller changes
(De Choudhury, Counts, Horvitz, CSCW 2013)
Measures % Acc. Prec. Recall F1 Spec. AUC
engagement 71.70% 0.757 0.708 0.727 0.742 0.734
Mean accuracy of extreme-change mothers (eng) 74.21%
emotion 71.21% 0.696 0.740 0.714 0.688 0.689
Mean accuracy of extreme-change mothers (emo) 70.94%
ego-net. 68.63% 0.725 0.797 0.757 0.510 0.633
Mean accuracy of extreme-change mothers (ego-net) 69.25%
style 72.90% 0.738 0.713 0.722 0.743 0.712
Mean accuracy of extreme-change mothers (style) 74.30%
Mean 71.11% 0.729 0.740 0.730 0.671 0.692
Std. dev. 0.023 0.032 0.020 0.075 0.075 0.032
Mean accuracy for predicting extreme changers 71.46%
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Counts, Horvitz, CSCW 2013)
(De Choudhury, Hoff, Rosaway, Horvitz, TechFest 2013)
understanding organizational behavior
Sharing affect-bearing content
(De Choudhury, Counts, CSCW 2013)
(De Choudhury, Counts, CSCW 2013)
understanding population behavior
(Sadilek et al, 2012)
(Fowler & Christakis, 2011)
o negative, low to moderate
activation moods are most
used
o a few positive, high
activation moods also show
considerable usage
o positive, moderate to high
activation moods tend to
be from more active
individuals (Lykken, 2000)
o negative, low and high
activation moods tend to
be used by fairly low
activity individuals
(De Choudhury, Counts, Gamon, ICWSM 2012)
Recall
Pre
cis
ion
joviality, fatigue, hostility and sadness are the best performing affects; guilt, shyness and attentiveness are the worst
(De Choudhury, Gamon, Counts, ICWSM 2012)
(Kicïman, De Choudhury, Rothschild, Gamon, Counts, Thiesson 2012)
we are attracted to social
media, in part due to
large-scale datasets
Is there something more
fundamental happening
here than just scale?
inferring meaningful social
ties in a network
(De Choudhury, Mason, Hofman, Watts, 2010)
@ Yahoo! Research
(De Choudhury, Lin, Sundaram, Candan, Xie, ICWSM 2010)
predicting information diffusion
in sampled networks
(De Choudhury, Counts, Czerwinski, ICWSM 2011)
Sampling high quality content
in social media leveraging
human cognition
the road ahead
o how can we motivate people to be more
conscious of the environment?
o how can social platforms be leveraged to help
under-privileged kids in education?