location and language in social media (stanford mobi social invited talk)

Stanford Mobi Social Workshop 2012 | Invited Talk!

Location and Language Use in Social Media!!

Ed H. Chi!!

Google Research!!Work done while at Palo Alto Research Center (Xerox PARC)!

!

!

2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk 1

What can you do with all this data?

Google Trends Trendalyzer

Big Data Analytics!! Google Analytics Google Website Optimizer


Model-Driven and Living Laboratory Approach

Characterization

and Modeling

Unlock Understanding

of Collective Intelligence

Intelligent UI

and Data-mining

Applications / Products Living Laboratory

Productization

3 2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk

2011-07-06 CSCL 2011 Keynote | Ed H. Chi

!"#$%&#'()*+,(+,(-#./(

2011-03-20 Adobe Distinguished Lecture 5

!#",012$(!"03$#(!  !"#$%&'"(")*+%,%-."

/&-01&0%#"21%-./34"5  6-0/*#7"5  8&-)&,*-"5  9&-.:-%#%"

!  ;-)"(":.7%,"/&-01&0%#"<&)/3="5  >&*?&-%#%"5  @%,A&-"5  B&$&-%#%"


415+,+6$#(7"08#"29#(:+912$,(2;"0,,(<219%29#(72""+#",(

B:*-."?:,'"?C"D*E7&-"F:-0G"@,%0:,*:"9:-H%,.*-:""IF:-0G"9:-H%,.*-:G"97*4"!9JK8"B1/3"LMNNO""

-0=+52=+01(>0"(:=%&?+19(<219%29#,(

!  >?*..%,"*#"&-"*-.%,-&.*:-&/"$7%-:A%-:-"5  8:#.",%#%&,E7"P:E1#%)":-"6-0/*#7"1#%,#"

5  Q1%#.*:-"&<:1."0%-%,&/*R&.*:-".:"-:-S6-0/*#7"

5  T-)%,#.&-)"E,:##S/&-01&0%"1#&0%")*+%,%-E%#"5  U%#*0-"*A$/*E&.*:-#"P:,"*-.%,-&.*:-&/"1#%,#"

!  V%#%&,E7"Q1%#.*:-#="5  J7&."*#".7%"/&-01&0%")*#.,*<1.*:-"*-">?*..%,W"

5  F:?"):"1#%,#":P")*+%,%-."/&-01&0%#"1#%">?*..%,W"

5  F:?"):"<*/*-01&/"1#%,#"#$,%&)"*-P:,A&.*:-"&E,:##"/&-01&0%#W"

"2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk 8

@2=2(A0$$#;=+01(B(!"0;#,,+19(

"NMX"/&-01&0%#""

MXCNYCNMSMZCN[CNM"\X"?%%'#]""

"[L8".?%%.#"

@::0/%"D&-01&0%";^!"_"D*-0^*$%"

">?*..%,"#.,%&A"

>:$"NM"/&-01&0%#"


)0C(DE(<219%29#,(+1()F+==#"(

((<219%29#( ((((()F##=,( ((((G( (((((H,#",(

I19$+,*( (NG`ZLG`[X" JD/D( ZGLYLG[Za"

K2C21#,#( NNG`aZGXL`" DL/D( NG((ZGMaX"

!0"=%9%#,#( ZG``(GZYX" L/M( ``(GMY("

41&01#,+21( (GXY(GYXL" J/M( ((YGNN["

:C21+,*( LG`(NGMLZ" N/O( aM[GZLL"

U1.E7" YY(G`XL" N4X" LXaGZL`"

b:,%&-" aZXGNY`" N4L" NN[GZM["

c,%-E7" [M(GaM[" N4M" L[NGXYN"

@%,A&-"" ZYYGXM`" N4M" N`LGXaa"

8&/&3" ZZ`G(YN" M4`" NYMGNXa"2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk 10

P%Q21RA0&+19(:=%&?(!  LGMMM",&-):A".?%%.#"P,:A"[L8".?%%.#"

!  L"71A&-"d1)0%#"P:,"%&E7":P".:$"N:"/&-01&0%#""5  -&.*H%"#$%&'%,#":,"$,:eE*%-."5  )*#E1##".:",%#:/H%")*#&0,%%A%-."

!  F&,)".:"e-)"!-):-%#*&-"_"8&/&3"d1)0%#"

!  ^,%#%-.%)"LGMMM".?%%.#".:"%&E7"d1)0%"

!  B1)0%"#%/%E.%)".?%%.#"*-"7*#C7%,"/&-01&0%"


-2;*+1#(5,/(P%Q21(

((<219%29#( ((((()R!( ((()RS( (((TRS( ((TR!( ((((((A0*#1U,(V2CC2(

I19$+,*( LON( LOD( WE( XJ( E/LJ(K2C21#,#( (aM" NGZ`Z" M" (Z" E/LN(!0"=%9%#,#( NaM" NGYM(" N`" Y" E/LW(41&01#,+21( NM[" NGYaZ" NZ" X" E/LD(:C21+,*( `[" NGYY`" NN" X" E/LW(@%=;*( NY" NGàY" L" L" E/LE(V0"#21( LX" NGà[" M" M" D/EE(c,%-E7" N(" NG`YM" M" a" M4a`"

@%,A&-"" NL" NGà`" L" a" M4aL"

8&/&3" Y" NGà`" X" `" M4ZZ"

)R!'(="%#(C0,+=+5#Y()RS'(="%#(1#92=+5#Y(TRS'(>2$,#R1#92=+5#Y(TR!'(>2$,#(C0,+=+5#(


Z;;%"2;?(0>(<219%29#(@#=#;=+01(

!  >?:">3$%#":P"6,,:,#"

5  !"#$%&$'(&)#$*+,-(.*$/%01,$23&'$4$#3#$3(#$%$#""-+"536#$'"(1,$768\)%.%E.%)"&#";P,*'&&-#]"

5  F*07"%,,:,",&.%"P:,".?%%.#":P"NfL"?:,)#"


-2;*+1#(5,/(P%Q21(

((<219%29#( ((((()R!( ((()RS( (((TRS( ((TR!( ((((((A0*#1U,(V2CC2(

c,%-E7" N(" NG`YM" E( O( M4a`"

@%,A&-"" NL" NG`a`" W( O( M4aL"

8&/&3" Y" NG`a`" N( L( M4ZZ"

•  c,%-E7="ZCa"cS^"7&H%"L"?:,)#"

•  @%,A&-="NCL"cSg"7&#"N"?:,)h"[Ca"cS^#"&,%"*-"6-0/*#7"

•  8&/&3="(CX"cSg#"_"aC`"cS^#"&,%"*-"!-):-%#*&-"


A0QQ01()F+==#"(A015#1=+01,(*2,*=29(

H[<(Q#1=+01(

"#C$?(\$%,S.?%%."A%.&)&.&]"

"#=F##=(2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk 15

H,#(0>(H[<,(+1(MW-()F##=,(((<219%29#( (H[<,(

!""# WDG(

6-0/*#7" LZi"

B&$&-%#%" DXG(

^:,.101%#%" DXG(

!-):-%#*&-" DXG(

K$&-*#7" NZi"

U1.E7" Nai"

b:,%&-" Nai"

c,%-E7" XOG(

@%,A&-"" XLG(

8&/&3" Nai"

!  A*+(:\%2"#(=#,=,(;013"Q#&(=*2=(&+]#"#1;#,(6?($219%29#(2"#(,+91+3;21=4(


:+91+3;21=(A"0,,R<219%29#(@+]#"#1;#,(((<219%29#( (H[<,( P2,*=29,( -#1=+01,( [#C$+#,( ([#=F##=,(

!""# WDG( DDG( NLG( XDG( DXG(

6-0/*#7" LZi" NXi" Xai" Lì" N(i"

B&$&-%#%" DXG( JG( X(i" ((i" OG(

^:,.101%#%" DXG( NLi" ZMi" (Li" NLi"

!-):-%#*&-" DXG( JG( OWG( WEG( XLG(

K$&-*#7" NZi" NNi" ZYi" (ì" NXi"

U1.E7" Nai" N(i" ZMi" (Zi" NNi"

b:,%&-" Nai" NNi" OXG( JLG( NNi"

c,%-E7" XOG( NLi" XYi" ([i" ì"

@%,A&-"" XLG( D^G( XMG( LZi" Yi"

8&/&3" Nai" JG( [Li" L(i" Lì"

A*+(:\%2"#(=#,=,(;013"Q#&(=*2=(&+]#"#1;#,(6?($219%29#(2"#(,+91+3;21=(


4QC$+;2=+01,(((<219%29#( (H[<,( (P2,*=29,( (-#1=+01,( ([#C$+#,( ([#=F##=,(

9//$ LNi" NNi" X`i" (Ni" N(i"

V0"#21( Nai" NNi" OXG( JLG( NNi"

_#"Q21"" XLG( D^G( XMG( LZi" Yi"

!  T#%":P">?*..%,"P:,"#:E*&/"-%.?:,'*-0"H#4"*-P:,A&.*:-"#7&,*-0")*+%,%-."*-")*+%,%-."/&-01&0%#"

!  U%#*0-":P",%E:AA%-)&.*:-"%-0*-%#"5  b:,%&-"1#%,#="$,:A:.%"E:-H%,#&.*:-&/".?%%.#"5  @%,A&-"1#%,#="$,:A:.%".?%%.#"?*.7"TVD#"


:=%&?+19(7+$+19%2$(7"08#",(!  !A$:,.&-E%":P"<,:'%,#"

5  K.,1E.1,&/"7:/%#"\j1,.k`L]G"D*H%B:1,-&/"\F%,,*-0"%."&/kMa]"

!  U%e-%"<*/*-01&/"<,:'%,#"&#"T#%,#"?7:".?%%.%)"*-"&"$&*,":P"/&-01&0%#"

!  9&H%&."

5  T-)%,S%#.*A&.%)")1%".:"XS?%%'".*A%"/*A*."

5  lH%,S%#.*A&.%)")1%".:"/&-01&0%")%.%E.*:-"%,,:,#"


S%Q6#"(0>(7+$+19%2$(7"08#",(E( J( P( I( S( D( K( F( G(

J( 140,730"

P( 488,545( 13,228"

I( 230,023" 4,825" 29,405"

S( 359,117" 10,139" 112,524" 36,068"

D( 150,041" 6,383" 30,855" 34,906" 30,916"

K( 19,722" 6,384" 906( 2,014" 1,109" 972(

F( 194,931" 10,463" 53,607" 34,586" 49,445" 33,568" 1,244"

G( 110,748" 6,053" 22,106" 21,471" 21,989" 22,162" 786( 24,763"

M( 148,365" 4,208" 31,184" 135,427" 31,967" 29,331" 1,518" 30,257" 18,301"


:*2"+19(H[<,(Z;"0,,(<219%29#,(E( J( P( I( S( D( K( F( G( M(

E 3,013" 18,399( 985" 4,986" 1,144" 212" 1,791" 1,647" 540"

J( 3,013" 77" 37" 58" 29" 43" 59" 46" 18"

P 18,399( 77" 74" 1,644" 198" 2" 453" 168" 123"

I( 985" 37" 74" 67" 64" 1" 53" 38" 279"

S 4,986" 58" 1,644" 67" 139" 0( 286" 139" 53"

D 1,144" 29" 198" 64" 139" 2" 112" 126" 48"

K 212" 43" 2" 1" 0( 2" 3" 3" 1"

F( 1,791" 59" 453" 53" 286" 112" 3" 157" 53"

G 1,647" 46" 168" 38" 139" 126" 3" 157" 40"

M 540" 18" 123" 279" 53" 48" 1" 53" 40"


:*2"+19(P2,*=29,(Z;"0,,(<219%29#,(

E( J( P( I( S( D( K( F( G( M(

E( 8,178" 33,197( 14,969"

27,284" 6,685" 798" 9,410" 7,208" 5,517"

J( 8,178" 331" 135" 351" 218" 149" 352" 260" 100"

P( 33,197( 331" 535" 4,682" 604" 13( 1,231" 580" 400"

I( 14,969" 135" 535" 762" 684" 25" 713" 415" 6,046"

S( 27,284" 351" 4,682" 762" 819" 28" 1,468" 708" 463"

D( 6,685" 218" 604" 684" 819" 26" 851" 769" 424"

K( 798" 149" 13( 25" 28" 26" 25" 18" 20"

F( 9,410" 352" 1,231" 713" 1,468" 851" 25" 879" 411"

G( 7,208" 260" 580" 415" 708" 769" 18" 879" 265"

M( 5,517" 100" 400" 6,046" 463" 424" 20" 411" 265"


4QC$+;2=+01,(!  !-)*E&.:,#":P"E:--%E.*:-"#.,%-0.7"<%.?%%-"

/&-01&0%#"5  g1A<%,":P"<*/*-01&/"<,:'%,#"5  ;E.#":P"<,:'%,&0%="#7&,*-0"TVD#"_"7&#7.&0#"

!  6-0/*#7"?%//"E:--%E.%)".:":.7%,#G"&-)"A&3"P1-E.*:-"&#"&"71<"

!  g%%)".:"*A$,:H%"E,:##S/&-01&0%"E:AA1-*E&.*:-#"

? �2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk 23

`<0;2=+01,a(:+912$,(+1(:0;+2$(-#&+2(

P#;*=Y(P019Y(:%*(B(A*+Y(AP4(WEDD(


b2"?+19(T0"Q,(0>(<0;2=+01(T+#$&()F+==#"(

5  c,%%SP:,A"5  91,,%-."/:E&.*:-"

_009$#(–  c,%%SP:,A"– 81/.*$/%"/:E&.*:-#"

T2;#6008(–  D*A*.%)":$.*:-#G""""""10(`72?(Z"#2Y(AZa(–  L"/:E&.*:-#"


Backstrom et al. 2010


Assumptions about the Location Field!

1.  Strongly-typed geo information!2.  Little noise!3.  Good precision!

29!2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk

27.7 million English Tweets (collected early

2010)

4.6 million Twitter Users

990K+ active Twitter users

10,000 Location Field Entries from Active

Twitter Users

Extracted Their Location Field

Entries

Randomly Sampled

10,000 Entries

Removed Automatically Populated Lat/

Lon Entries (1154)

8846 Manually Entered Twitter Location Field Entries


31

Two Coders Powered by human knowledge, the Internet, friends + family, etc.


89%+ Agreement

2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk

Some Valid Geographic Information

66%

Nothing Entered

18%

Non-Valid Geographic!Information!

16%!

Study 1: “Geographicness”!

“850 n.benson ave! upland ca”!

“JoviLand, CA”!

“San Francisco”!

“the panhandle”!

“Middle Earth”!

“Global Citizen”!

data quality of the location field!

32!

“New Mexico”!

“Novi Sad, Serbia, Europe”!

“The Moon”!

“Worldwide”!

“kcmo – call the popo”!


Information Type # of Users

Popular Culture Reference 195 (12.9%)

Privacy-Oriented 18 (1.2%)

Insulting or Threatening to Reader 69 (4.6%)

Non-Earth Location 75 (5.0%)

Negative Emotion Towards Current Location 48 (3.2%)

Sexual in Nature 49 (3.2%)

Study 1: Non-Geo Information types of non-geographic information entered into the location field









types of non-geographic information entered into the location field

Study 1: Non-Geo Information


“BieberTown”

“My World”

“belieber wonderland”

“JaeJoongs heart”

“Next to Waldo :D”

“somewhere in Glambertville”

“Los Angeles, 2019 (GET IT?)”

“Schrute Farms”

Study 1: Popular Culture References Non-geographic information in the location field in user’s profiles









Study 1: Non-Geo Information types of non-geographic information entered into the location field


“Stalker City”

“Stalking me here isnt enough?”

“MindingMyOwn”

“For me to know n u to find out”

“NONE YA BISNESS”

“UM…STALKER!!”

“kgb answers”

Study 1: Privacy References Non-geographic information in the location field in user’s profiles


Study 1: Implications

Geocoder

Latitude and Longitude Coordinates

STRONGLY-TYPED GEOGRAPHIC INFORMATION

REQUIRED


Study 1: Quality Implications

Geocoder

Latitude and Longitude Coordinates

16% Non-Valid Geographic

Information

STRONGLY-TYPED GEOGRAPHIC INFORMATION

REQUIRED


Study 1: Quality Implications

Yahoo! Geocoder

Non-Valid Geographic Information

16%

“Stalker City”, “NONE YA BISNESS”, “Justin Biebers Heart”, “The Void”, “Redneck Hell”, “In the Middle of Nowhere”, “yer mum”, “BSNBC”, “in God’s Graces’, etc…


41

“Loserville :)” (-71.397524, 42.28904)


42

“With God” (19.13683,47.705132)


43

“Justin Biebers heart!” (-91.700189, 36.328785)


Some Valid Geographic Information

!

66%


27.7 million English Tweets (collected early

2010)

4.6 million Twitter Users

990K+ active Twitter users

10,000 Location Field Entries from Active

Twitter Users

Extracted Their Location Field

Entries

Randomly Sampled

10,000 Entries

Removed Automatically Populated Lat/

Lon Entries (1154)



33.687456,-84.244945 Seriously?

1154 users?


Position Information

Self-reported Sensor-based

Global Positioning System (GPS)

WiFi Access Point

Cell Phone Towers

Implicitly Revealed!


72.10% Accuracy 2.91x better than random

United States? Canada?

United Kingdom? Australia?

Study 2: Country Experiments Uniform Sampling

49

Naïve B

ayes C

lassifier


United States? Canada?

United Kingdom? Australia?

88.86% Accuracy 1.08x better than random

Study 2: Country Experiments Demographically Proportional Sampling

50

Naïve B

ayes C

lassifier


Study 2: State Experiments

Naïve B

ayes C

lassifier

30.28% Accuracy

California? Arkansas? New York?

Washington? Texas?

…

5.45x better than random

Uniform Sampling


Study 2: State Experiments

Naïve B

ayes C

lassifier

27.31% Accuracy

California? Arkansas? New York?

Washington? Texas?

…


Demographically Proportional Sampling


Word Geography Predictiveness calgary Canada 419.42 brisbane Australia 137.29 coolcanuck Canada 78.28 afl Australia 56.24 clegg UK 35.49 cbc Canada 29.40 yelp USA 19.80

Study 2: Predictive Words

Word Geography Predictiveness elk Colorado 90.74 redsox Massachusetts 41.18 biggbi Michigan 24.26 gamecock South Carolina 16.00 crawfish Louisiana 14.87






Tweets Have Implicit Location Information


This needs to be considered in the

context of implicit location

disclosure!


Contributions!

1. First characterization study of user location field behavior !

2. Location field behavior is much more complex than has been assumed!

3. The complexity has implication for geography-related HCI technologies!

4. Location field behavior must be considered along with implicit disclosure behavior.!

!

56!2012-04-04 Stanford Mobi Social Workshop 2012 Invited Talk

:0;+2$(:="#2Q([#,#2";*(!  ;-&/3.*E#"

5  c&E.:,#"*A$&E.*-0",%.?%%.&<*/*.3"IK17"%."&/G"!666"K:E*&/"9:A$1.*-0"LMNMO"

5  D:E&.*:-"e%/)":P"1#%,"$,:e/%#"IF%E7."%."&/G"9F!"LMNNO"5  c"921+;(dBZ(6#*25+0",(e!2%$(#=(2$Y(4Af:-UDDg(

5  <219%29#,(%,#&(+1()F+==#"(eP019(#=(2$Y(4Af:-UDDg(

!  !A$,:H*-0"K.,%&A"6m$%,*%-E%"

5  >:$*ES<&#%)"#1AA&,*R&.*:-"_"<,:?#*-0":P".?%%.#"Ij%,-#.%*-"%."&/G"T!K>LMNMO(

5  >?%%.",%E:AA%-)&.*:-"I97%-"%."&/G"9F!LMNM"_"9F!LMNNO"


H,#"(7#*25+0"(+1(:0;+2$(-#&+2(!  T#%,#"&):$."&-)"&)&$."#:E*&/"A%)*&".:"#1*.".7%*,"-%%)#4"!  U%#*0-*-0"P:,".7%"H&,*%.3":P"<%7&H*:,"*#"E,*.*E&/".:".7%"

#1EE%##":P"#:E*&/"A%)*&4"!  J7&."?%"/%&,-%)="

5  D:E&.*:-"e%/)"E&-"<%"n1*.%"%m$,%##*H%h"5  D&-01&0%"&+%E.#"1#%"E&#%#h"5  D&-01&0%"E&-"<%"&"<&,,*%,".:"%m$,%##*:-"&-)"*-P:,A&.*:-"

<,:'%,&0%4"


)*218(?0%h(!  E7*o&EA4:,0"!  7..$=CC%)E7*4-%."


location and language in social media (stanford mobi social invited talk)

Business

stanford mobi social

invited talk

social media

datamining productization

keynote ed

language use

adobe distinguished