Qagela ubani? Izibonelo ezi-5 zokuthi kungani ukususa amagama akuyona inketho
Isingeniso ku-Guess Who
Qagela ubani? Yize ngineqiniso lokuthi iningi lenu liyazi lo mdlalo kusuka emuva ezinsukwini, nazi impinda emfushane. Umgomo womdlalo: thola igama lomlingiswa wekhathuni okhethwe umphikisi wakho ngokubuza imibuzo ethi 'yebo' no 'cha', njengokuthi 'ingabe umuntu ufaka isigqoko?' noma 'umuntu ufaka izibuko'? Abadlali basusa abakhethiweyo ngokuya ngempendulo yomphikisi futhi bafunde izimfanelo ezihlobene nomlingiswa oyimfihlakalo wabaphikisi babo. Umdlali wokuqala obala umlingiswa oyimfihlakalo womunye umdlali uwina umdlalo.
Uyitholile. Umuntu kufanele akhombe umuntu ngaphandle kwedathasethi ngokuba nokufinyelela kuphela kuzimpawu ezihambisanayo. Eqinisweni, sihlala siwubona lo mqondo we-Guess Who wafaka isicelo, kepha wasebenza kumasethi edatha afomethwe ngemigqa namakholomu aqukethe izimfanelo zabantu bangempela. Umehluko omkhulu lapho usebenza nedatha ukuthi abantu bathambekele ekubhekeni phansi ubulula abantu bangempela abangadalulwa ngabo ngokuthola izimfanelo ezimbalwa kuphela.
Njengoba umdlalo we-Guess Who ukhombisa, othile angakhomba abantu ngokufinyelela kuzimpawu ezimbalwa kuphela. Kusebenza njengesibonelo esilula sokuthi kungani kususwa 'amagama' kuphela (noma ezinye izikhombi eziqondile) kudathasethi yakho kwehluleka njengenqubo yokufihla amagama. Kule bhulogi, sinikezela ngamacala amane asebenzayo ukukwazisa ngezingozi zobumfihlo ezihambisana nokususwa kwamakholomu njengendlela yokwenza igama lingaziwa.
2) Ukuhlaselwa kokuxhumanisa: idathasethi yakho exhunywe kweminye imithombo yedatha (yomphakathi)
Ubungozi bokuhlaselwa kokuxhumana yisizathu esibaluleke kunazo zonke sokuthi kungani ukususa amagama kungasebenzi (akusasebenzi) njengendlela yokwenza ungaziwa. Ngokuhlaselwa kokuxhumeka, umhlaseli uhlanganisa idatha yoqobo neminye imithombo yedatha efinyelelekayo ukuze akhombe ngokuhlukile umuntu othile futhi afunde imininingwane (evame ukubucayi) ngalo muntu.
Okusemqoka lapha ukutholakala kweminye imithombo yedatha ekhona manje, noma engaba khona ngokuzayo. Cabanga ngawe. Ingakanani imininingwane yakho yomuntu siqu engatholakala ku-Facebook, Instagram noma ku-LinkedIn engahle isetshenziselwe ukuhlukunyezwa ngokuhlaselwa kokuxhumana?
Ezinsukwini ezedlule, ukutholakala kwemininingwane bekukhawulelwe kakhulu, okuchaza ukuthi kungani ukususwa kwamagama kwakwanele ukugcina ubumfihlo babantu. Idatha engatholakali kancane isho amathuba ambalwa wokuxhumanisa idatha. Kodwa-ke, manje singabahlanganyeli (abasebenzayo) emnothweni oqhutshwa yidatha, lapho inani ledatha likhula ngezinga lomchazi. Imininingwane eminingi, nokwenza ngcono ubuchwepheshe bokuqoqa idatha kuzoholela ekwandeni kokuhlaselwa kokuxhumana. Yini umuntu angayibhala eminyakeni eyishumi mayelana nengozi yokuhlaselwa ngokuxhumana?
Umzekeliso 1
Idatha ekhula ngokwandayo iyiqiniso
Ucwaningo lwesigameko
USweeney (2002) ukhombise ephepheni lezemfundo ukuthi ukwazile kanjani ukuthola nokuthola idatha ebucayi yezokwelapha kubantu abathile ngokuya ngokuxhumanisa isethi yedatha etholakalayo yomphakathi 'yokuvakashelwa esibhedlela' kumbhalisi wokuvota otholakala emphakathini e-United States. Kokubili amasethi edatha lapho kucatshangwa ukuthi angaziwa kahle ngokususwa kwamagama nezinye izikhombi eziqondile.
Umzekeliso 2
Ukuhlaselwa kokuxhumeka ekusebenzeni
Ngokuya ngemingcele emithathu kuphela (1) i-Zip Code, (2) Ubulili kanye (3) Nosuku Lokuzalwa, ukhombise ukuthi ama-87% abo bonke abantu baseMelika angaphinda akhonjwe ngokufanisa izici ezichazwe ngenhla kuzo zombili izinqolobane zedatha. USweeney wabe esephinda umsebenzi wakhe wokuba 'nezwe' njengenye indlela 'ye-Zip Code'. Ngaphezu kwalokho, ukhombisile ukuthi i-18% yabantu bonke base-US ingakhonjwa kuphela ngokufinyelela kudathasethi equkethe imininingwane mayelana (1) nezwe lasekhaya, (2) ubulili kanye (3) nosuku lokuzalwa. Cabanga ngemithombo yomphakathi eshiwo ngenhla, njenge-Facebook, i-LinkedIn noma i-Instagram. Ngabe izwe lakho, ubulili nosuku lokuzalwa kuyabonakala, noma abanye abasebenzisi bayakwazi ukulikhipha?
Umzekeliso 3
Imiphumela kaSweeney
| Izinkomba ze-Quasi | % ekhonjwe ngokuhlukile kubantu base-US (izigidi ezingama-248) |
| I-ZIP enamadijithi ama-5, ubulili, usuku lokuzalwa | 87% |
| indawo, ubulili, usuku lokuzalwa | 53% |
| izwe, ubulili, usuku lokuzalwa | 18% |
Lesi sibonelo sikhombisa ukuthi kungaba lula ngokumangazayo ukungachazi abantu kudatha ebonakala ingaziwa. Okokuqala, lolu cwaningo lubonisa ubukhulu obukhulu bengozi, njengoba Ama-87% wabantu base-US angabonakala kalula kusetshenziswa izici ezimbalwa. Okwesibili, imininingwane yezokwelapha eveziwe kulolu cwaningo ibibucayi kakhulu. Izibonelo zedatha yabantu abadaluliwe kusuka kudathasethi yokuvakasha esibhedlela zifaka phakathi ubuhlanga, ukuxilongwa kanye nemithi. Izimfanelo umuntu angakhetha ukuzigcina ziyimfihlo, ngokwesibonelo, ezinkampanini zomshuwalense.
3) Abantu abanolwazi
Enye ingozi yokususa okokuhlonza okuqondile kuphela, njengamagama, ivela lapho abantu abanolwazi benolwazi oluphakeme noma ulwazi mayelana nezici noma ukusebenza kwabantu abathile kudathasethi. Ngokuya ngolwazi lwabo, umhlaseli angabe esekwazi ukuxhumanisa amarekhodi wedatha athile kubantu bangempela.
Ucwaningo lwesigameko
Isibonelo sokuhlaselwa kwedathasethi usebenzisa ulwazi oluphakeme yicala lamatekisi laseNew York, lapho i-Atockar (2014) ikwazile ukudalula abantu abathile. Idathasethi eqashiwe ibiqukethe lonke uhambo lwamatekisi eNew York, olunothile ngezimpawu eziyisisekelo ezifana nezixhumanisi zokuqala, izixhumanisi zokuphela, intengo nethiphu yohambo.
Umuntu onolwazi owazi ukuthi iNew York ukwazile ukuthola uhambo lwamatekisi aya kwiklabhu yabantu abadala 'iHustler'. Ngokuhlunga 'indawo yokugcina', wehlise amakheli okuqala ngqo futhi ngaleyo ndlela wakhomba izivakashi ezahlukahlukene ezifika njalo. Ngokunjalo, umuntu angadonsa ukugibela kwamatekisi lapho ikheli lasekhaya lomuntu lowo laziwa. Isikhathi nendawo kasaziwayo abambalwa bama-movie batholwa kumasayithi ezinhlebo. Ngemuva kokuxhumanisa lolu lwazi nedatha yamatekisi e-NYC, bekulula ukuthola abagibeli babo bamatekisi, inani abalikhokhile, nokuthi ngabe sebetholile yini.
Umzekeliso 4
Umuntu onolwazi
idrophu iqondisa uHustler
UBradley Cooper
Jessica Alba
4) Idatha njengezigxivizo zeminwe
Umugqa ojwayelekile wokuphikisana uthi 'le datha ayinalutho' noma 'akekho noyedwa ongenza noma yini ngale datha'. Lokhu kuvame ukuba umbono ongaqondile. Ngisho nedatha engenacala kakhulu ingakha 'isigxivizo somunwe' esiyingqayizivele futhi isetshenziselwe ukukhomba umuntu ngamunye. Kuyingozi etholakala ekukholweni ukuthi idatha uqobo ayinalutho, kuyilapho kungenjalo.
Ubungozi bokuhlonza buzokhula ngokwanda kwedatha, i-AI, namanye amathuluzi nama-algorithms avumela ukutholakala kobudlelwano obuyinkimbinkimbi kudatha. Ngenxa yalokho, noma ngabe idathasethi yakho ingenakutholakala manje, futhi kungenzeka ukuthi ayisebenzi kubantu abangagunyaziwe namuhla, kungenzeka ingabi kusasa.
Ucwaningo lwesigameko
Isibonelo esihle lapho kwenzeka khona ukuthi iNetflix ihlose ukuthola umnyango wayo we-R & D ngokwethula umncintiswano ovulekile weNetflix wokuthuthukisa uhlelo lwabo lokuncoma ama-movie. 'Leyo ethuthukisa i-algorithm yokuhlunga ngokubambisana ukubikezela izilinganiso zomsebenzisi zamafilimu iwina umklomelo we-US $ 1,000,000'. Ukuze isekele isixuku, i-Netflix ishicilele idathasethi equkethe kuphela izimfanelo eziyisisekelo ezilandelayo: i-userID, i-movie, usuku lwebanga kanye nebanga (ngakho-ke alukho olunye ulwazi ngomsebenzisi noma ifilimu uqobo).
Umzekeliso 5
Isakhiwo sedatha yedatha yeNetflix
| I-ID Yomsebenzisi | I-Movie | Usuku lwebanga | Grade |
| 123456789 | Mission Impossible | 10-12-2008 | 4 |
Ngokuzihlukanisa, idatha ibonakale iyize. Lapho kubuzwa umbuzo othi 'Ngabe kukhona imininingwane yamakhasimende kudathasethi okufanele igcinwe iyimfihlo?', Impendulo ibithi:
Cha, yonke imininingwane ekhomba amakhasimende isusiwe; konke okusele izilinganiso nezinsuku. Lokhu kulandela inqubomgomo yethu yobumfihlo… '
Kodwa-ke, uNarayanan (2008) wase-University of Texas e-Austin ukufakazele okunye. Inhlanganisela yamamaki, usuku lwebanga ne-movie yomuntu ngamunye yakha i-movie-fingerprint eyingqayizivele. Cabanga ngokuziphatha kwakho kwe-Netflix. Ucabanga ukuthi bangaki abantu ababuke isethi efanayo yama-movie? Bangaki ababuke isethi efanayo yama-movie ngasikhathi sinye?
Umbuzo oyinhloko, ungamatanisa kanjani nale zigxivizo zeminwe? Kwakulula kunalokho. Ngokuya ngemininingwane evela kwiwebhusayithi eyaziwa kakhulu yokulinganisa ama-movie IMDb (Internet Movie Database), kungakhiwa isigxivizo seminwe esifanayo. Ngenxa yalokho, abantu bangaphinda bachazwe.
Ngenkathi isimilo sokubuka ama-movie singahle singathathwa njengolwazi olubucayi, cabanga ngokuziphatha kwakho - ungakhathazeka uma kungaziwa emphakathini? Izibonelo uNarayanan azinikezile ephepheni lakhe yizintandokazi zepolitiki (izilinganiso ku-'Jesu waseNazaretha 'kanye' neVangeli likaJohane ') nezintandokazi zocansi (izilinganiso ku-'Bent' kanye ne-'Queer as folk ') ezingasuswa kalula.
5) Umthethonqubo Wokuvikelwa Kwemininingwane Jikelele (GDPR)
I-GDPR ingahle ingathokozisi kakhulu, noma ichashazi lesiliva phakathi kwezihloko zebhulogi. Noma kunjalo, kuyasiza ukuthola izincazelo ziqonde lapho ucubungula imininingwane yomuntu. Njengoba le bhulogi imayelana nombono oyiphutha ovamile wokususa amakholomu njengendlela yokwenza idatha ingaziwa nokuthi ikufundise njengeprosesa yedatha, ake siqale ngokuhlola incazelo yokufihla amagama ngokuya kwe-GDPR.
Ngokuya nge-recital 26 evela ku-GDPR, imininingwane engaziwa ichazwa njenge:
'imininingwane engahlobene nomuntu okhonjiwe noma ongakhonjwa wemvelo noma idatha yomuntu siqu engaziwa ngomuntu ngendlela yokuthi leyo ndaba ayaziwa noma ayisabonakali.'
Njengoba eyodwa icubungula imininingwane yomuntu ephathelene nomuntu wemvelo, yingxenye 2 yencazelo kuphela efanelekile. Ukuze uhambisane nencazelo, umuntu kufanele aqinisekise ukuthi isihloko sedatha (umuntu ngamunye) asibonakali noma asisabonakali. Njengoba kukhonjisiwe kule bhulogi, noma kunjalo, kulula kakhulu ukukhomba abantu abathile ngokuya ngezimpawu ezimbalwa. Ngakho-ke, ukususa amagama kudathasethi akuhambisani nencazelo ye-GDPR yokufihla amagama.
Ekuphetheni
Siphonsele inselelo indlela eyodwa ebhekwe njengokujwayelekile futhi, ngeshwa, isasetshenziswa njalo indlela yokwenza ukuthi igama lingaziwa: ukususa amagama. Kumdlalo we-Guess Who nezinye izibonelo ezine mayelana:
- Ukuhlaselwa kokuxhumanisa
- Abantu abanolwazi
- Idatha njengezigxivizo zeminwe
- Umthetho Wokuvikelwa Kwaziswa Okujwayelekile (GDPR)
kukhonjisiwe ukuthi ukususa amagama kwehluleka njengokufihla amagama. Yize izibonelo zingamacala esiteleka, ngasinye sikhombisa ubulula bokuphinda kukhonjwe kanye nomthelela omubi ongaba khona kubumfihlo babantu.
Ekuphetheni, ukususwa kwamagama kudathasethi yakho akuholeli kudatha engaziwa. Ngakho-ke, kungcono sigweme ukusebenzisa womabili la magama ngokungafani. Ngithemba ngeqiniso ukuthi ngeke usebenzise le ndlela yokwenza ungaziwa. Futhi, uma usaqhubeka, qinisekisa ukuthi wena nethimba lakho niziqonda ngokugcwele izingozi zobumfihlo, futhi nivunyelwe ukwamukela lezo zingcuphe egameni labantu abathintekile.
Idatha iyenziwe, kodwa ithimba lethu lingokoqobo!
Xhumana noSyntho futhi omunye wochwepheshe bethu uzoxhumana nawe ngesivinini sokukhanya ukuze ahlole inani ledatha yokwenziwa!
- D. Reinsel, J. Gantz, uJohn Rydning. Ukudijithali Komhlaba Kusuka ku-Edge kuye ku-Core, i-Data Age 2025, 2018
- L. Sweeney. k-ukungaziwa: imodeli yokuvikela ubumfihlo. Ijenali Yomhlaba Wonke Ngokungaqiniseki, Ubuxhakaxhaka Nohlelo Olususelwe Olwazini, 10 (5), 2002: 557-570
- L. Sweeney. Izibalo Ezilula Zabantu Zivame Ukukhomba Abantu Ngokuhlukile. ICarnegie Mellon University, Iphepha Lokusebenza Lobumfihlo beDatha 3. IPittsburgh 2000
- P. Samarati. Ukuvikela Ubunikazi Babaphenduli Ekukhishweni KweMicrodata. Ukuthengiselana kwe-IEEE ku-Knowledge and Data Engineering, 13 (6), 2001: 1010-1027
- I-Atockar. Ukuhamba nezinkanyezi: Ubumfihlo Bomgibeli kwi-NYC Taxicab Dataset, ngo-2014
- UNarayanan, A., noShmatikov, V. (2008). Ukukhishwa ngamandla kwe-robust de database enkulu ye-sparse. Ku-Proceedings - 2008 IEEE Symposium on Security and Privacy, SP (kk. 111-125)
- I-General Data Protection Regulation (GDPR), i-Recital 26, Ayisebenzi Emininingwane Engaziwa