2.4.1.3 Censorship ea metswedi ya dikgang ka ho 'muso o Chinese

Bafuputsi fata Chinese tsoang sechabeng mecha ea litaba liwebsaete ho ithuta li thibela. Ba ileng a sebetsana le incompleteness le latent-tšobotsi diplomat.

Ho phaella ho data ya tse khōlō le sebelisitsoe ka mehlala e 'meli e fetileng, bafuputsi ba ka boela bokella bona ya data hae observational, joalokaha ho ne ho hlollang bontšoa ke Gary King, Jennifer Pan,' me Molly Roberts ' (2013) lipatlisiso ka hlahlobang lingoliloeng ka' muso Sechaena.

Social posts by mecha ea phatlalatso ea Chaena ba hlahlojoa ke e sabaletseng e mmuso didirisiwa e nahana ho kenyeletsa ba likete tse mashome batho. Bafuputsi le baahi ho le joalo, na ka kutloisiso e nyenyane ea kamoo censors tsena etsa qeto ea seo dikahare lokela ho hlakolwa ho tloha metswedi ya dikgang. Litsebi tse of China ha e le hantle ba le litebello tse hanyetsanang eo mefuta eohle ea lintho tse ngotsoeng moo ke ka ho fetisisa ka etsahala hore ebe ho hlakolwa. Ba bang ba nahana hore censors tsepamisa kelello posts by tse mahlonoko tseo ho leng a puso ea ha ba bang ba nahana hore ba ameha ka lintho tse ngotsoeng moo o khothalletsang boitshwaro ja kopanetsweng, tse kang ea boipelaetso. Qalang ho ithuta e leng oa litebello tsena e nepahetse na tshepo bakeng sa kamoo bafuputsi utloisisa Chaena le mebuso e meng ea bompoli hore kopanela hlahlobang lingoliloeng. Ka lebaka leo, Morena le 'basebetsi-'moho ba ne ba batla ho bapisa posts by e neng e hatisitsoeng' me hamorao o hlakolwa ho posts by e neng e hatisitsoeng me le ka mohla hlakolwa.

Ho bokella lintho tse ngotsoeng moo tsena amehang hlollang boenjiniere ea hahaba ho feta 1,000 Chinese tsoang sechabeng mecha ea litaba websites-hore e mong le fapaneng leqepheng mapatlelong-ho fumana posts by tshwanetseng, 'me joale revisiting posts by tsena ho bona tse neng li o ile hlakolwa. Ho phaella ho mathata boenjiniere ho tloaelehile amahanngoa le ka tekanyo e kholo web-gagaba, morero ona o ne a talimane le phephetso phaella ka ho re e hlokahalang e le hore ba itima lijo ka ho fetisisa hobane ba bangata ba hlahlojoa posts by ba isoa fatše ka lihora tse ka tlaase ho 24. Ka mantsoe a mang, e leng crawler liehang ne a hopotse lotho ea posts by e neng e hlahlojoa. Ho feta moo, crawlers ile ka tlameha ho etsa sena pokello data yohle ha a ntse a tlohe phumano esere sa kahisano mecha ea phatlalatso ea websites thibela ho fihlella kapa ho seng joalo fetola maano a bona e le karabelo ea thuto.

Ka lekhetlo le leng sena sa setonanahali boenjiniere mosebetsi o ile oa phethoa, Morena le 'basebetsi-'moho ba ne ba fumana ba ka bang limilione tse 11 posts by ka lihlooho 85 fapaneng tse neng li pele ho boletsweng thehiloeng ho bophahamo ba bona a se lebeletse ho nahanela. Ka mohlala, sehlooho ea nahanela ea phahameng ke Ai Weiwei, moo moetsi dissident; sehlooho ea nahanela bohareng ke kananelo ho bile ho devaluation ea chelete Chinese, le taba ea nahanela tlaase ke Cup ea Lefatše. Ea posts by tsena limilione tse 11 ba ka bang limilione 2 neng ba hlahlojoa, empa posts by ka lihlooho mamelang le e ikemiselitseng nkeloa holimo li ne li hlahlojoa hanyenyane feela hangata ho feta posts by ka lihlooho bohareng le tlaase nahanela. Ka mantsoe a mang, censors Chinese ba ka bang jwalo ka ha ho ka etsahala hore e hlahlobe a morago ga kalogo e buang Ai Weiwei joalokaha poso e bua ka senoelo ea Lefatše. fumanoeng ke tsena ba ne ba sa lumellane le khopolo bebofatsa taba eo 'muso o censors posts by tsohle ka lihlooho hlokolosi.

Sena se manolotsoeng bonolo hakana le sekhahla li thibela ka ho sehlooho ka ba khelosa Leha ho le joalo. Ka mohlala, 'muso o ka' na hlahlobe posts by tse tsehetsang Ai Weiwei, empa tlohela posts by tse mahlonoko tseo ho leng sa hae. E le hore ho khetholla pakeng tsa lintho tse ngotsoeng moo hloko haholoanyane, bafuputsi ba lokela ho lekanya maikutlo ka mong poso. Kahoo, tsela e 'ngoe ho nahana ka eona ke hore maikutlo ka mong poso a sa bohlokwa latent tšobotsi ea e mong le poso. Ka bomalimabe, ho sa tsotellehe mosebetsi o mongata, mekhoa itirisang tsa ka botlalo ya phumano maikutlo sebelisa lidikishinari pele ho le teng ba ntse ba ha e ntle haholo le maemo a mangata (nahana ka mathata a ho bōpa e lobaka lo longwe maikutlong September 11, 2001 ho tloha Karolo 2.3.2.6). Ka lebaka leo, Morena le 'basebetsi-'moho ho hlokahala tsela ea ho nka bona ba limilione tse 11 metswedi ya dikgang posts by mabapi le hore na ba ne ba 1) mahlonoko tseo ho leng a puso ea, 2) tšehetsa puso, kapa 3) irrelevant kapa e thehiloeng' neteng litlaleho tse mabapi lintho tse etsahalang. Sena se utloahala eka mosebetsi setonanahali, empa ba rarolloa e sebelisa leqheka matla; ngoe eo ho tloaelehile ya data saense ea bitsoang empa hona joale ha ho bapisoa e leng ka seoelo haholo saenseng ea sechaba.

Taba ea pele, a talima mohato wa tloaelo bitsoa pele ho e lokisa, bafuputsi sokoloha lipalong metswedi ya dikgang ka har'a motsoako oa popeho fumanoa tokomane e telele, moo ho neng moleng o le mong bakeng sa e mong le e fumanoa tokomane le kholomo ya e 'ngoe eo a tlalehiloeng hore na poso e-na le lentsoe tse tobileng tseo (mohlala, ipelaetsa, sephethephethe, jj.) Ka mor'a moo, sehlopha sa bathusi ba etsa lipatlisiso ka letsoho ngotsweng ya maikutlo ea le mehlala feela ea poso. Joale, Morena le 'basebetsi-'moho le sebelisoa ena letsoho ngotsweng ya data ho hakanya mochine ithuta ea mohlala e neng e ka fihlela qeto ea sa maikutlo ea poso e thehiloeng litsobotsi tsa eona. Qetellong, ba ile ba sebelisa mochine ena ithuta ea mohlala ho ba hakanya ea maikutlo ea posts by tsohle tse limilione tse 11. Kahoo, ho e-na bowena ho bala le ho bo bitsa posts by limilione tse 11 (e leng ne e tla ba ke ke ha khoneha logistically), ba ka bowena ngotsweng palo e nyenyane ea posts by 'me joale a sebelisa seo ya data-rasaense ba ne ba bitsa ithuta o tsamaisoa ho hakanya mekgahlelo ya posts by tsohle. Ka mor'a hore ho tlatsa Analysis ena, Morena le 'basebetsi-'moho ba ile ba khona ho etsa qeto ea hore tsela e batlang e ho makatse hore ebe kgonego ya poso e ntse hlakolwa ne amaneng le hore na ho ne ho nyatsa ba mmuso kapa tšehetsa mmuso.

Figure 2.3: KHATISO tlhohlomiso bakeng sa Tsamaiso e sebelisitsoeng King, Pan, and Roberts (2013) ho ka go fopholetsa ea maikutlo ea Chinese posts by limilione tse 11 tsa phedisano mecheng ea litaba. Taba ea pele, a talima mohato wa tloaelo bitsoa pele ho e lokisa, bafuputsi sokoloha sa kahisano mecha ea phatlalatso ea posts by ka har'a motsoako oa popeho fumanoa tokomane e telele (sheba Grimmer and Stewart (2013) bakeng sa boitsebiso bo eketsehileng). Ea bobeli, bafuputsi letsoho coded ea maikutlo ea le sampole e nyenyane ea posts by. Ea boraro, bafuputsi ba koetlisitsoeng e laola tsamaiso ithuta ea mohlala ho bee sehlopheng sa maikutlo ea posts by. Ntlha ea bone, bafuputsi ba sebelisa o tsamaisoa ho ithuta ea mohlala ho ba hakanya ea maikutlo ea posts by tsohle. Sheba King, Pan, and Roberts (2013) , Sehlomathiso sa B bakeng sa tlhaloso e qaqileng ho feta.

Qetellong, Morena le 'basebetsi-'moho fumana hore mefuta e meraro feela ea posts by ne ba kamehla hlahlojoa: litšoantšo tsa bootsoa, ​​nyatsa censors,' me ba neng kopanelo ea bokgoni ba (khr, monyetla oa ho ea isang ho ho ipelaetsa ka tekanyo e kholo). Ka ho sheba palo e ngata ea posts by e neng e hlakolwa le lintho tse ngotsoeng ba neng ba sa hlakolwa, Morena le 'basebetsi-'moho ba ile ba khona ho tseba hore na censors sebetsa feela ke shebeletse' me bala. A etsa lipatlisiso morago ga moo e, ba ile ba ha e le hantle ka ho toba o ile a kenella ho kena ka har'a Chinese tsoang sechabeng mecha ea litaba bottler ke bōpa posts by le dikahare hlophisehileng fapaneng le ho lekanya e leng fumana hlahlojoa (King, Pan, and Roberts 2014) . Re tla ithuta ho eketsehileng ka le atamela liteko Khaolong 4. Ho ekelletsa moo, ho neng ho tsoantsetsa sehlooho tse tla etsahala ho pholletsa buka eo, mathata ana-e leng latent-ise khanya diplomat ka linako tse ling ho rarolla le ne le laola tsamaiso ithuta-retelehela ho ba tloaelehileng haholo ka ho etsa lipatlisiso tsa phedisano ka dijethale lilemo tsa ka. U tla bona litšoantšo tšoana haholo le mokhabo 2.3 Khaolong ea 3 (Ho botsa lipotso) le 5 (Bōpang tshebedisano boima); ho ke ke e mong oa mehopolo 'maloa tseo hlaha ka likhaolo multiple.

Ohle a mararo ka mehlala-e leng tsena boitshwaro ja sebetsa ea bakhanni tekesi New York, setsoalle sebopeho ke liithuti, 'me metswedi ya dikgang li thibela boitšoaro ba Machaena muso e bontša joang hore batlang e le bonolo baloa ha ya data observational ka etsa hore bafuputsi ba hore a leke tse boletsoeng esale pele mogopolofela. Maemong a mang, ya data khōlō nolofalletsa hore u etse tshephile ena e batlang e ho toba (joaloka tabeng ea New York Taxis). Maemong a mang, bafuputsi tla hloka ho bokella ba bona ba ya data hae observational (joaloka tabeng ea hlahlobang lingoliloeng Chinese); sebetsana le incompleteness ke kopanngwe ya data ba phutheha mmoho (joaloka tabeng ea ya netweke hore lintho li iphetotse); kapa ho etsa mofuta o itseng oa latent-tšobotsi diplomat (joaloka tabeng ea hlahlobang lingoliloeng Chinese). Ha ke ntse ke le tšepo ea mehlala ena e bontša, etsoe babatlisisi ba khonang ho botsa lipotso tse thahasellisang, tse kholo le tšepiso e khōlō.