5.3.1 Netflix Prize

Netflix Prize wuxuu isticmaalaa call u furan tahay in la saadaaliyo oo movies dadka jeclaan doonaa.

Mashruuca furan call ugu fiican u yaqaan waa Prize Netflix ah. Netflix waa shirkadda online kirada movie, iyo in 2000 waxaa la bilaabay Cinematch, adeeg inay ku talinaynaa movies macaamiisha. Tusaale ahaan, Cinematch aragto laga yaabaa in aad jeclaan Star Wars iyo weerarada Empire Back ka dibna ku talinaynaa in aad daawato Return of Jedi ah. Markii hore, Cinematch liidata shaqeeyay. Laakiin, in ka badan koorsada ah sannado badan, Cinematch sii si loo hagaajiyo awoodda in la saadaaliyo waxa movies macaamiisha ku raaxaysan lahaa. By 2006, si kastaba ha ahaatee, horumarka on Cinematch plateaued. The cilmi ee Netflix ayaa isku dayay wax kasta oo badan, waxay u malaynayaan kari, laakiin waqti isku mid ah, ay la tuhunsan yahay in ay jiraan fikrado kale oo caawin kara iyaga horumariyaan nidaamka. Sidaas darteed, waxay la iman jireen waxa uu ahaa, waqtiga, xal xagjirka ah: call furan.

Halis u ah guusha ugu danbeyn ee Prize Netflix waxay ahayd sida call bannaan waxaa loogu talagalay, oo design this waxay leedahay casharo muhiim u ah sida wicitaanada furan loo isticmaali karaa cilmi-baarista bulshada. Netflix ma ma ahan oo keliya saaray codsi qaabaysnayn fikrado, taas oo ah waxa dad badan qiyaasi marka ugu horeysa ka fiirsan call furan. Saas ma aha, Netflix wadaan dhibaato cad qodobada a fudud qiimeynta: waxay loolan dadka si ay u isticmaalaan set oo ah 100 million ratings movie in la saadaaliyo 3 million ratings qaban-out (ratings in dadka isticmaala sameeyey laakiin in Netflix ma sii daayo). Qof kasta oo la abuuro laga yaabaa in isku geynta ah in la saadaaliyo kara 3 million ratings qaban-baxay 10% ka wanaagsan tahay Cinematch ku guuleysan doono 1 milyan oo doolar. Tani waxay si cad oo sahlan si aad u codsato qiimaynta shuruudaha-barbar ratings saadaaliyay in ay qaban-baxay ratings-loogu tala galay in Prize Netflix la haystaa oo sidan oo kale in xal sahlan si aad u hubiso ka badan dhalin ah; waxay u leexatay caqabad ku ah hagaajinta Cinematch galay dhibaato ku haboon call furan.

Bishii October ee 2006, Netflix sii daayay dataset ka kooban 100 million ratings movie ka ku saabsan oo ku saabsan 500,000 macaamiisha (waxaan ka fiirsan doonaa saameynta gaarka ah ee saamaxaaddu xogta this Cutubka 6). Xogta Netflix la conceptualized karaa sida shaxda weyn in qiyaastii waa 500,000 macaamiisha by 20,000 filimada. Matrix this gudahood, waxaa jiray 100 million ratings on qiyaasta ah 1 ilaa 5 stars (Shaxda 5.2). Hawsha ugu weyn ay ahayd in ay isticmaalaan xogta lagu arkay furta in la saadaaliyo 3 million ratings qaban-baxay.

Shaxda 5.2: Schematic xogta ka Prize Netflix ah. Netflix sii daayay oo ku saabsan 100 milyan oo ratings (1 xiddiga in 5 stars) ay bixiyaan 500,000 oo macaamiil ah oo ku saabsan 20,000 filimada. Ujeedada Prize Netflix waxay ahayd in ay isticmaalaan ratings si saadaaliyaan ratings qaban-out of 3 million movies, muujiyay sida "?". ratings saadaaliyay soo gudbiyeen ka qaybgalayaasha in Prize Netflix ah ayaa marka la barbar dhigo ratings qaban-baxay. Waxaan ka hadli doonaa arrimaha anshaxa wareegsan sii daayo macluumaadka this Cutubka 6.
movie 1 movie 2 movie 3 . . . movie 20,000
Customer 1 2 5 . ?
Customer 2 2 ? . 3
Macaamiisha 3 ? 2 .
. . . . . . . .
Customer 500,000 ? 2 . 1

Cilmi iyo Anonymous oo dunida ku ahaayeen inay caqabad ku tahay, oo 2008 in ka badan 30,000 oo qof ayaa ka shaqeeya waxa on (Thompson 2008) . In ka badan koorsada ee tartanka, Netflix ka badan 40,000 oo xal la soo jeediyay in ka badan 5,000 oo kooxo helay (Netflix 2009) . Sida iska cad, Netflix akhriyi kari waayeen oo garta, kuwaas oo dhan xal soo jeediyay. Waxa oo dhan si habsami orday, si kastaba ha ahaatee, sababtoo ah waxay ahaayeen xal sahlan tahay in la hubiyo. Netflix kaliya yeelan karaan kombiyuutar isbarbar ratings saadaaliyay in ay ratings qaban-soo ton (ton gaar ahaa xididka square qalad hoose-gees ay isticmaali) a pre-cayimay. Waxa uu ahaa awoodda this si dhakhso ah loo qiimeeyo xal in karti loo Netflix inay aqbalaan xal ka qof kasta, taas oo soo baxday in muhiim sababtoo ah fikradaha wanaagsan ka qaar ka mid ah meelaha la yaab leh yimid. Dhab ahaantii, xal guusha waxaa soo gudbiyeen koox bilaabay saddex cilmi in uu lahayn nidaamka talo movie aragnimo dhismaha ka hor (Bell, Koren, and Volinsky 2010) .

Mid ka mid ah dhinaca quruxda badan ee Prize Netflix waa in ay qof kasta oo dunida awood u leedahay in ay xal cadaalad qiimeeyo. Marka dadku uploaded ratings ay saadaaliyay, iyagu ma ay u baahan tahay in geliyaan aqoonsiga ay tacliinta, da'dooda, jinsiyad, jinsiga, nooca galmada, ama wax kasta oo ku saabsan naftooda. Sidaas darteed, waxay ratings saadaaliyay of professor caan ka Stanford ayaa la daweeyey dhab ahaan la mid ah sida kuwa ka da'yar oo iyada jiifka. Nasiib darro, tani run in cilmi bulshada ugu ma aha. Taasi waa, waayo cilmi bulshada ugu, qiimaynta waa waqti aad u gubaya oo qayb ahaan shakhsi ahaan. Sidaas daraaddeed, fikrado cilmi ugu yihiin marnaba si dhab ah qiimeeyaa, oo markii fikradaha la qiimeeyaa, way adag tahay in fullaacin qiimaynta kuwa ka abuuraha fikradahaas. Maxaa yeelay, xal waa ay fududahay in la hubiyo, wicitaanada furan ogolaan cilmi si ay u helaan oo dhan xal laga yaabo cajiib ah in dhici lahaa iyada oo dildilaaca haddii ay tixgeliyaa kaliya xal ka Professor caan ah.

Tusaale ahaan, at hal dhibic inta lagu guda jiro qof Netflix Prize magaca shaashadda Simon Funk posted on inuu blog xal soo jeediyay ku salaysan tahay kala daatay a qiimaha keli, hab ka algebra toosan oo aan horay loo isticmaalo by qaybgalayaasha kale. post blog Funk ee ahaa isku mar farsamo iyo weirdly rasmi. Oo post blog this ku tilmaamay xal fiican ama waxa ay ahayd waqti lumis? Ka baxsan mashruuc call furan, xalka laga yaabaa in marnaba heshay qiimeyn halis ah. Ka dib oo dhan Simon Funk ma ahaa borofisar ka tirsan Cal Tech ama MIT, wuxuu ahaa sameeysa barnaamijyada software-ka, kuwaas oo, waqtiga, waxaa backpacking agagaarka New Zealand (Piatetsky 2007) . Haddii uu saxafadeed fikrad this si injineer ah ee Netflix, waxaa hubaal ma loo qaaday si dhab ah.

Nasiib wanaag, sababtoo ah shuruudaha qiimaynta ahaayeen cad oo sahlan si aad u dalbato, ratings uu saadaaliyay ayaa qiimeeyaa, oo ay ahayd goobtii cad in uu hab ahaa mid aad u xoog badan, ayuu rocketed inay kaalinta afaraad ee tartanka, natiijo wayn siiyey in kooxo kale oo hore u ahayd u shaqeeya bilood on dhibaatada. In dhamaadka, qaybo ka mid ah hab Simon Funk ayaa la isticmaalay by gebi ahaanba tartanka halis ah (Bell, Koren, and Volinsky 2010) .

Xaqiiqada ah in Simon Funk doortay in ay qoraan post blog a sharraxaad ka hab, halkii ay isku dayaya in ay sii qarsoodi ah, sidoo kale waxay muujinaysaa in ay ka qaybgalayaashu badan ee Prize Netflix ma ay si gaar-gasha abaalmarinta million dollar. Saas ma aha, ka qaybgalayaasha badan ayaa sidoo kale u muuqdeen in ay ku raaxaystaan ​​oo caqabad ku ah garaadka iyo bulshada in la horumariyo agagaarka dhibaatada (Thompson 2008) , dareenka aan ka filan cilmi badan fahmi karin.

Netflix Prize waa tusaale classic of call furan. Netflix wadaan su'aal gool gaar ah (saadaalinta ratings movie) iyo xal ka dad badan baryo. Netflix ahaa awoodaan si ay u qiimeeyaan xal kuwaas oo dhan, maxaa yeelay, waxay ahaayeen kuwa fudud si loo xaqiijiyo ka badan si ay u abuuraan, iyo ugu dambeyntii Netflix gaaray xalka ugu wanaagsan. Next, waxaana idin tusin doonaa sida habkani isku loo isticmaali karaa bayoolaji iyo sharciga.