hou commentary

Keia pauku, ua papahana e, e hoʻohana 'e like me ka olua, i ole e heluhelu ia ka moolelo.

• Introduction (Pauku 2.1)

Kekahi ano o ka malama ana i ka mea i komo ole iloko o keia mokuna he ethnography. No ka mea, nui ma ethnography ma kikohoʻe hakahaka ike Boellstorff et al. (2012) , a no ka mea, nui ma ethnography i ka haumāna kikohoʻe, a me ke kino hakahaka ike Lane (2016) .

• Big ikepili (Pauku 2.2)

I ka wa a oukou e repurposing aeaiiua, aia he elua noʻonoʻo 'ia paena e hiki e kokua mai ia oukou e oe i halawai me ka hiki pilikia hoʻomaopopo. Mua, e hiki ia ia ke hoao aku, e noonoo i ka maikaʻi dataset no kou pilikia, a me ka hoohalike ai i ke dataset ka mea au e hoʻohana 'ana. Pehea ka poe like, a pehea la heʻokoʻa lakou? Inā 'oe i ohi i ko oukou mau ikepili oe ia oe iho, o ka poe e kiʻi mai ia e likeʻole ma waena o ka mea a oukou i makemake a me ka mea a oukou i loaa mai. Aka, he e hooholo ina keia oko ao he Nā hana a nui.

Elua, no ka mea e hai i hana, a ohi i ko oukouʻikepili no kekahi kumu. Oe e ho'āʻo e hoomaopopo i ko lakou hoopaapaa ana. I keia ano o ka aʻe, 'enekinia, ke kōkua mai iāʻoe e hōʻike ana hiki pilikia a me na biases iloko o ko oukou repurposed aeaiiuo.

Ka mea, aole hookahi kona manaʻo ma ka ho'ākāka 'ana o ka "nui ikepili", akā, he nui wehewehe he e kālele ana ma luna o ke 3 ia makou: (e like me, ka leo,ʻano, a me ka māmā holo Japec et al. (2015) ). Ma kahi o ka hoʻomōhala ma na ano o ka 'ike, kuu ka ho'ākāka' ana e kālele nui ma ke kumu o ka ikepili i hanaia.

I koʻu hoʻi o ke Aupuni hoʻomalu 'ikepili i loko o ka mahele o ka nui ike, he iki unusually. Kekahi poe i hana i keia hihia, komo Legewie (2015) , Connelly et al. (2016) , a me Einav and Levin (2014) . No ka mea, e pili ana i ka waiwai o ke aupuni hoʻomalu ike no ka noiʻi, ike Card et al. (2010) , Taskforce (2012) , a me Grusky, Smeeding, and Snipp (2015) .

No ka mea, o ka manao o ka hoʻomalu noiʻi, mai loko mai o ke aupuni ana helu nenoaiu, pakahi aku la ia i ka US helu Buro, ike Jarmin and O'Hara (2016) . No ka mea, o ka buke loa lapaau o ka hoʻomalu moʻolelo noiʻi ma ka 'ikepili helu Kuekene, ike Wallgren and Wallgren (2007) .

Ma ka mokuna, I komo hoʻohālike au i kahiko ana e like me ka General Social Survey (GSS) i ka nohona Media aeaiiuo kumu e like me Twitter. No ka mea, i ka hoʻokolokolo pono a me ka akahele e like ma waena o kahiko ana, a me ka nohona me Media aeaiiua, ike Schober et al. (2016) .

• Na ano o ka nui ikepili (Pauku 2.3)

Keia mau 10 'ano o ka nui ike, ua oleloia ua hiki aku ma keʻano o naʻano likeʻole ma keʻano o ke kākau okoa. Palapala i alakaʻi ai i koʻu manaʻo i kēia mau nīnūnē ioiinyony: Lazer et al. (2009) , Groves (2011) , Howison, Wiggins, and Crowston (2011) , boyd and Crawford (2012) , Taylor (2013) , Mayer-Schönberger and Cukier (2013) , Golder and Macy (2014) , Ruths and Pfeffer (2014) , Tufekci (2014) , Sampson and Small (2015) , Lewis (2015) , Lazer (2015) , Horton and Tambe (2015) , Japec et al. (2015) , a me Goldstone and Lupyan (2016) .

Ma keia Mokuna, Fashion hoohana i ka makahiki kikohoʻe ko läkou ', ka mea au i manao, kŘpa aʻAkika nā. Kekahi kaulana manawa no kikohoʻe ko läkou 'ua kikohoʻe footprints (Golder and Macy 2014) , aka, e like me Hal Abelson, Ken Ledeen, a Harry Lewis (2008) kuhikuhi aku, he oi aku kūpono kau mea paha kikohoʻe fingerprints. I ka wa e ho okumu i footprints, ua maopopo na mea i hanaia, a me ko oukou mau footprints hiki ole nui e Hahai I ia oukou kino. Ka ia, aole oiaio no ko oukou kikohoʻe ko läkou '. I ka mea, ua haalele ko läkou 'a pau i ka manawa ana i ia oe i mea uuku ike. A, eia nae keia mau ko läkou 'aʻole' oe i kou inoa maluna o lakou, aole lakou e hiki pinepine e ua hoʻopili hou aku ai ia oukou. Ma na olelo e, ka poe nui e like me fingerprints: invisible a kino nä.

Big

No ka mea, nui no ke aha nui datasets, hoike ana helu ho'āʻo ma problematic, ike Lin, Lucas, and Shmueli (2013) , a me McFarland and McFarland (2015) . Keia mau kumuhana e alakai noiʻi e kālele ana ma ka mea hiki significance ma mua o ana helu significance.

Mau-ma

I ka wa e manao mau-ma aeaiiuo, he mea nui e noonoo i ko oukou ua kapakai a ka kiko'ī ia poʻe kānaka ma luna o ka manawa, a ina paha oukou e hoʻohālikelikeʻia kekahi huli hui o na kanaka; ike no ka laʻana, Diaz et al. (2016) .

Ole-reactive

A Classic buke i ole-reactive ana o Webb et al. (1966) . Na ano he kumu hoʻohālike i loko o ka buke mālama '-la ka mīkini makahiki, aka, ua nō ka mālamalama. No ka mea, examples o na kanaka iki i ko lakou hana no o ke alo o ka nuipa a hoʻomakākiu, ike Penney (2016) , a me Brayne (2014) .

kāpili

No ka mea, nui ma ka moolelo linkage, ike Dunn (1946) , a me Fellegi and Sunter (1969) (Land) a Larsen and Winkler (2014) (kēia). Similar aku la, ua ulu nō hoʻi i loko o kekahi polokalamu kamepiula, 'epekema malalo o na inoa e like me ka aeaiiuo deduplication, manawa' ike, ka inoa'ālike, nānā me ka loaʻaʻana, ae nānā aku ka loaʻaʻana (Elmagarmid, Ipeirotis, and Verykios 2007) . He nui no hoi pilikino ka hoopakele ana i hoʻokokoke mai, e palapala linkage i ole koi i ka EYI o kino nä 'ike (Schnell 2013) . no hoi, ua ulu Facebook i ke hoomau aku i ka loulou i ko lakou mau mooolelo, i ke koho 'ana; ua hanaia keia e loiloi i ka hoʻokolohua ka mea aʻu e hai aku ia oukou i ka Mokuna 4 (Bond et al. 2012; Jones et al. 2013) .

No ka mea, hou ma? Ieoaeunoai validity, ike Shadish, Cook, and Campbell (2001) , Mokuna 3.

Inaccessible

No ka mea, hou maluna o ka AOL huli log debacle, ike Ohm (2010) . I kaumaha aʻo e pili ana aia i ka poe a me na aupuni ma ka Mokuna 4 i ka wa aʻu e wehewehe hoʻokolohua. A helu o na mea kākau, ua kaheaia pili ana noiʻi ka mea nui maluna o inaccessible ikepili, ike Huberman (2012) , a me boyd and Crawford (2012) .

Hana maikai ala no kulanui noiʻi e loaʻa iā 'ike ke kōkua o ka mea e hana i kekahi poʻe e like me ka intern a me ke kipa kanaka noiʻi. A iâ lâkou e ho'ā aeaiiuo ke kōkua o, keia kaʻina e hoi kōkua i ke kanaka noiʻi aʻo hou e pili ana i ua hana i ka 'ikepili, ka mea nui no Ka Ikepili.

Ole-elele

Ole-representativeness ka mea, he nui no na pilikia noiʻi a me na aupuni e makemake e hoike ana i ka heluna kanaka holoʻokoʻa. Keia mea emi mai o ka manao no ka poʻe a pau i kāu hoʻomōhala maluna o ko lakou mau users. No ka mea, hou ma luna o ka 'ikepili helu Netherlands E hoomanao i ke kahe o ka ole-representativeness o ka hana nui aeaiiua, ike Buelens et al. (2014) .

Ma ka Mokuna 3, au e hōʻike mea hōʻikeʻikeʻuʻuku a me ka manaʻo i loko o ka oi aku mamuli. A hiki inaʻikepili he ole-elele, malalo o kekahi kanawai, aole lakou e hiki ke kaumaha i ka paka maikai koho.

naulu la

System 'Aeʻa he nui e ike mai i ke waho. Akā naʻe, i na MovieLens papahana (kūkākūkā hou i loko o ka Mokuna 4) Ua holo aku no ka oi mamua o 15 makahiki e kekahi me kēia noiʻi hui. Nolaila, aole lakou i nä waihona palapala a pili 'ike e pili ana i ke ala o ka nenoaiu i Hawaiʻi ma luna o ka manawa, a pehea la keia i ka hopena Ka Ikepili (Harper and Konstan 2015) .

A helu o na haumana, ua hoʻomōhala ma 'Aeʻa ma Twitter: Liu, Kliman-Silver, and Mislove (2014) , a me Tufekci (2014) .

Algorithmically hoʻohilahilaʻia

I mua lohe i ke kau "algorithmically hookahuli" hoʻohana 'ia e Jon Kleinberg ma ka olelo ana. I ka papa kuhikuhiE manaʻo ma hope o performativity oia i kekahi nohona nauka theories he "kiko'ī, aole Panel" (Mackenzie 2008) . Oia hoi, lakou nae i 'ano o ke ao nei ma mua pono hoʻopio ia mea.

Dirty

Hui ana helu lapaʻau nāna e kāhea aku aeaiiuo, i ka hoʻomaʻemaʻe, ana helu aeaiiuo hoʻoponopono. De Waal, Puts, and Daas (2014) hōʻike ana helu ikepili hoʻoponopono 'ana i kūpono hoʻomohala no ka anamanaʻo ka ikepili a me ka hoike aku i ka mea i lakou, ua pili i ka nui ikepili kumu, a me na Puts, Daas, and Waal (2015) makana kekahi o ia manaʻo no ka oi nui hoolohe mai.

No ka mea, he mau laʻana o nā haʻawina hoʻomōhala ma ka leka uila ma Twitter, Clark et al. (2016) , a me Chu et al. (2012) . Eia ke oki, Subrahmanian et al. (2016) wehewehe i nā hualoaʻa o ka DARPA Twitter Bot FIRST.

ikepili koʻikoʻi

Ohm (2015) Waikiki mamua noiʻi ma luna o ka manaʻo o ka 'ikepili' ike, a kaumaha he nunui-ololi hōʻike. Ka eha kumumea oia i hāpai 'ia no, ke probability o ka poino; probability o ka poino; imua o kekahi anamanaʻo pili; a ina paha o ka pilikia noonoo majoritarian pili.

• Helu mea (Pauku 2.4.1)

Farber ka like ana o taxis ma New York, ua nānā 'ana i ka mua o? Anoee ma Camerer et al. (1997) i hoʻohana ekoluʻokoʻa pono Eia kekahi laʻana o ka pepa hele a lahilahi-pepa 'ano apau loa i hoʻohanaʻia ma Keaukaha, e palapala huakaʻi hoʻomaka manawa, hope manawa, a me ka momona. Keia mua like loaa i makemake Keaukaha e lilo i pale umauma hoʻokahi earners; lakou hana e emi i na la a ko lakou uku, he kiekie ae.

Kossinets and Watts (2009) Ua hoʻomōhala i nā kêia kanaka a ke homophily i lawelawe latike. Ike Wimmer and Lewis (2010) no ka mea, he okoa e hoʻokokoke aku i ka mea pilikia i hoʻohanaʻikepili mai Facebook.

Ma mahope hana, ke alii, a me nā hoapili i hoi kaupaona i nā online, censorship ma Kina (King, Pan, and Roberts 2014; King, Pan, and Roberts 2016) . No ka mea, ke hai e hoʻokokoke aku ana online, censorship ma Kina, ike Bamman, O'Connor, and Smith (2012) . No ka mea, e oi aku maluna o ana helu epekema e like me ka mea lawelawe i loko o King, Pan, and Roberts (2013) ke hoʻomaopopoʻia ka olelo la o ka 11 miliona kia, ike Hopkins and King (2010) . No ka mea, ua oi maluna o ke hoʻoponopono i ke aʻo, ike James et al. (2013) (emi oaoieei-) a me Hastie, Tibshirani, and Friedman (2009) (oi oaoieei-).

• Forecasting (Pauku 2.4.2)

Forecasting mea he nui loa o ka hanalima ike 'epekema (Mayer-Schönberger and Cukier 2013; Provost and Fawcett 2013) . Kekahi 'ano o ka forecasting i ua mau hana e lawelawe noiʻi, ua HI forecasting, no ka mea hoʻohālike Raftery et al. (2012) .

Google Flu mea kūʻo ia i ka mua papahana e hoʻohana i nā 'ike e nowcast influenza prevalence. In mea, noiʻi i loko o ka United States (Polgreen et al. 2008; Ginsberg et al. 2009) , a me Kuekene (Hulth, Rydevik, and Linde 2009) ua loaa ia ia kekahi mau hua'ōlelo (e like me, "flu") wānana ke aupuni lehulehu ola hoʻomakākiu aeaiiua ma mua o ka mea, ua hookuu. A laila nui, nui no hoi na hana i hoao ai e hoʻohana mīkini kumumeaʻikepili no ka maʻi hoʻomakākiu ka loaʻaʻana,ʻike Althouse et al. (2015) i ka moʻolelo loiloi.

A iâ lâkou e hoʻohana mīkini kumumea ikepili e kilokilo ola nā haumāna, he mea no hoi i ka nui nui o ka hana me Twitter kaʻikepili e kilokilo koho nā haumāna; no ka mea, hōʻike manaʻo ike Gayo-Avello (2011) , Gayo-Avello (2013) , Jungherr (2015) (Ch. 7), a me Huberty (2015) .

E ho ohana i huli aeaiiua i ka luaÿi influenza prevalence, a me ka hoʻohana 'ana Twitter kaʻikepili e kilokilo ke koho, he elua ano he kumu hoʻohālike no ka hoʻohana kekahi ano o ka mīkini kumumea e kilokilo i kekahi ano o ka hanana i loko o ke ao nei. Malaila mea nui kēia helu o nā haʻawina a pau i kēia mau 'ole. Papa 2,5 nā he kakaikahi na examples.

• Approximating ho okolohua (Pauku 2.4.3)

Ka puke pai PS Kalaiaina Science i ka symposium ma nui ikepili, causal kuhi, a me ka olelo kumumanaʻo, a me Clark and Golder (2015) manaʻoi hoʻopōkole 'kela makana. Ka puke hana o ka National Ke Kai O Sciences o ka Amerika Huipuia i kekahi symposium ma causal kuhi, a nui ka ikepili, a Shiffrin (2016) manaʻoi hoʻopōkole 'kela makana.

I ka olelo o maoli hoʻokolohua, Dunning (2012) hoʻolako he maikaʻi puke loa lapaau. No ka mea, nui ma ka hoʻohana 'ana i ka Vietnam kikoo lottery e like me ka maoli hoʻokolohua, ike Berinsky and Chatfield (2015) . No ka mea, mīkini aʻo hoʻokokoke mai i hoao mai ae e koho wehe maoli ia mea i loko o ka nui ikepili kumu, ike Jensen et al. (2008) , a me Sharma, Hofman, and Watts (2015) .

I ka olelo o ka'ālike, no ka mea, he optimistic manual, ike Stuart (2010) , a no ka pessimistic manual ike Sekhon (2009) . No ka mea, hou maluna o'ālike e like me ka ano o ka pahi, ike Ho et al. (2007) . No ka mea, na buke i hoomakaukau ohe. Maikaʻi o ka'ālike, ike Rosenbaum (2002) , Rosenbaum (2009) , Morgan and Winship (2014) , a me Imbens and Rubin (2015) .