Algorithm ka mor'a 'mino

Suthela holimo, American Idol.

Pale e latelang e kholo ea katleho indastering ea 'mino e ke ke ea fumanoa litlholisanong tsa talenta tse phahameng. Ho e-na le hoo, e tla khetholloa ka lihlopha tsa data ka li-algorithms tse rarahaneng tse etselitsoeng ho senola ts'ebeliso le mekhoa ea khoebo.

Ka holim'a metsi, mokhoa ona o utloahala o omme ebile o se na maikutlo ho feta litlhaloso tsa Simon Cowell, empa ha e le hantle ke eona tsela ea ho qetela eo sechaba se khethang "ntho e kholo e latelang." Nako le nako ha sechaba se tobetsa lihokelo tsa YouTube, se beha linepe tsa konsarete ho Twitter, kapa li qoqa ka lihlopha ho Facebook, li kenya letsoho ho sehlopha sa tlhahisoleseling se bitsoang data e kholo. Lentsoe lena le bolela pokello ea lisebelisoa tsa data tse kholo le tse nang le likamano tse rarahaneng. Nahana ka sebopeho sa marang-rang a marang-rang. Li na le limilione tsa litlaleho tsa basebelisi ka bomong tse hokahanngoang hammoho ka setsoalle, ‘lintho tseo ba li ratang’, litho tsa sehlopha, joalo-joalo.Ha e le hantle, liipone tse kholo tsa data sebopeho sa lipolanete tsena.

Indastering ea 'mino, boitsebiso bo bongata bo hlahisoa ke mesebetsi e kang thekiso ea inthaneteng, ho khoasolla, le likhokahano tse etsoang ka li-app kapa libaka tsa mecha ea litaba ea sechaba. Lipalo-palo tse lekantsoeng li kenyelletsa "nako ea linako tseo lipina li bapaloang kapa li fetisitsoeng ka tsona, hammoho le boemo ba ho hohela ha tsona mecheng ea litaba tsa sechaba ho ipapisitsoe le liketso tse kang tseo Facebook li li ratang le li-tweets." Lisebelisoa tsa ho hlahloba li khetholla ho tsebahala ka kakaretso ha maqephe a fan le ho ngolisa maikutlo a nepahetseng kapa a fosahetseng ka baetsi ba litšoantšo. Ka kopanelo, boitsebiso bona bo qolla tse tumileng ha jwale, bo lekola dijithale tsa bataki, mme bo lebisa dithekisong ka batho ba se nang balekane, thepa, dithekethe tsa dikhonsarete, esita le ho ingodisa bakeng sa ditshebeletso tsa ho bapalla mmino.

Mabapi le ho sibolla talente e ncha, lintlha tse kholo li phetha karolo ea bohlokoa ho hlahiseng thahasello lileiboleng tse kholo tsa rekoto. Maemong a mangata, lik'hamphani li fana ka maikutlo a leqephe la sebini, 'like', le balateli. Joale, lipalo li ka bapisoa habonolo khahlanong le litsebi tse ling tsa mofuta o tšoanang. Hang ha ketso e se e hlahisitse balateli ba likete tse lekholo hammoho le Facebook kapa Twitter, batsamaisi ba litalenta ba ela hloko 'me ba qala ho hlahisa thahasello ka har'a indasteri ea' mino ka boeona.

Lintlha tse kholo tse khethang tse latelang tse kholo tsa Top 40

Bokhoni ba ho khetholla mekhoa ea hona joale le ho bolela esale pele megastar e latelang e tla le meputso e meholo ea lichelete bakeng sa bohle ba amehang. Ka mohlala, boitseanape ba boitsebiso ba ithutile ka phello ea mecha ea litaba ea sechaba ho albamo ea iTunes le ho latela thekiso ka ho bapisa metrics ea motho le lekeno la ba bang. Ba fihletse qeto ea hore ts'ebetso ea mecha ea litaba ea sechaba e amana le keketseho ea thekiso ea lialbamo le lipina. Haholo-holo, maikutlo a YouTube a na le tšusumetso e kholo ho thekiso; phumano e entseng hore batho ba bangata ba rekotiloeng ba kenye livideo tsa 'mino tse kholo tsa tekanyetso sethaleng ho khothaletsa batho ba sa nyaloang. Pele o sebelisa limilione ho hlahisa livideo, tlhahlobo e sebelisoa ho tseba hore na ke lipina life tse ka bang monate ho ipapisitse le mesebetsi ea inthaneteng ea bamameli ba shebiloeng. Ho nepahala ha likhakanyo tsena ho amana le boleng ba tlhahlobo e kholo ea data.

Bo-rakhoebo ba ka har'a indasteri ea 'mino ba se ba ntse ba leka mekhoa e mecha ea ho hlahisa mekhoa e metle ea ho kotula tlhahisoleseding ka katleho e kholo le ho nepahala. O mong oa mehlala e hlokomelehang ke khoebo e kopanetsoeng lipakeng tsa EMI Music le Data Science London e bitsoang The EMI Million Interview Dataset. E hlalosoa e le "e 'ngoe ea li-database tse ruileng ka ho fetisisa le tse kholo ka ho fetisisa tsa kananelo ea' mino tse kileng tsa fumaneha - pokello ea lintlha tse kholo, tse ikhethang, tse ruileng, tsa boleng bo holimo tse kopantsoeng ho tsoa liphuputsong tsa lefats'e tse nang le lithahasello, maikutlo, boitšoaro, tloaelo le kananelo ea 'mino joalo ka ha ho boletsoe ke barati ba mmino.”

David Boyle, Motlatsi e Moholo oa Insight ho EMI Music, oa hlalosa, "(E) e na le lipuisano tse limilione tse buang ka lihlooho tse kang boemo ba takatso ea mofuta o itseng oa 'mino le mofuta o mong, mekhoa e ratoang ea ho sibolla' mino, libini tse ratoang tsa 'mino, menahano ea bosholu ba 'mino, ho phalla ha' mino, liforomo tsa 'mino le palo ea balateli. ”

Sepheo sa morero ke ho tsebisa sechaba pokello ena ea litaba le ho ntlafatsa boleng ba khoebo indastering ea 'mino.

Boyle o re: "Re bile le katleho e kholo ea ho sebelisa lintlha ho thusa rona le litsebi tsa rona ho utloisisa bareki, 'me re thabela ho arolelana tse ling tsa lintlha tsa rona ho thusa ba bang ho etsa se tšoanang," ho bolela Boyle.

Ka 2012, EMI Music and Data Science London e ile ea nka morero mohato o mong hape ka ho amohela Music Data Science Hackathon. EMC, moetapele oa lefats'e oa saense ea data le litharollo tse kholo tsa data, o ile a ikopanya le khoebo mme a fana ka lisebelisoa tsa IT. Ka nako ea lihora tse 24, bo-ramahlale ba data ba 175 ba ile ba qapa mekhoa e 1,300 XNUMX le mekhoa ea ho araba potso e reng: “Na u ka bolela esale pele hore na momameli o tla rata pina e ncha?” Liphetho li ile tsa fana ka maikutlo a matla a bohlale bo kopanetsoeng 'me barupeluoa ba ile ba qapa mekhoa e hlalositsoeng e le ea maemo a lefatše.

"Litlhaloso tse senotsoeng ka hackathon ena li fana ka maikutlo a matla le bokhoni boo Big Data bo nang le bona - ka bobeli bakeng sa ho sibolloa ha kelello le bakeng sa boleng bo eketsehileng ba khoebo bakeng sa mekhatlo ea mefuta eohle," ho bolela Chris Roche, Mookameli oa Libaka oa EMC Greenplum.

Empa u lefa baetsi ba litšoantšo joang?

Ka mor'a hore indasteri e etse qeto ea hore pina e na le bokhoni ba ho e lokolla e le 'ngoe, e bala meputso joang ha pina e bapaloa lipapaling tsa litaba tsa sechaba kapa libaka tsa marang-rang? Hona joale, "litlaleho tsa lirekoto tsa boholo bohle li tobane le bothata bo ntseng bo hola ba ho hokahanya lintlha tse tsoang lik'hamphaning tse tsamaisang litaba tse kang Spotify, Deezer le YouTube, empa ho na le batho ba fokolang ho feta leha e le neng pele ho etsa joalo."

E 'ngoe ea liqholotso tse ka sehloohong ho tloha ponong ea tsamaiso ea tlhahisoleseding ke hore mekhoa e mengata ea tsamaiso ea polokelo ea boitsebiso e ne e sa ntlafatsoa ho sebetsana le lihlopha tsa data tse kholo le tse rarahaneng joaloka data e kholo. Mohlala, boholo ba lifaele tsa data tsa dijithale tse hlahisoang ke baphatlalatsi ba mmino bo feta hole kamoo mananeo a kang Excel a ka sebetsanang le ona. Sena se baka mathata ho kenyelletsa data e sieo le lileibole tsa lifaele tse sa tsamaellaneng le software ea accounting.

Maemong a mangata, litaba tsena kaofela li hlophisoa ke li-accounting, ho eketsa nako le mosebetsi o eketsehileng mosebetsing o seng o ntse o le boima. Maemong a mangata, karolo e kholo ea chelete ea li-label e tlameletsoe lefapheng la likarabello.

Ho loants'a litaba tsena, bo-rakhoebo ba theha liforomo tsa bohlale ba khoebo tse nang le bokhoni ba ho hlophisa le ho sekaseka lintlha tse kholo. E 'ngoe ea mehlala e metle ka ho fetisisa ke k'hamphani ea Austria ea Rebeat, e hlalosang litšebeletso tsa bona e le "accounting ea borena ka ho tobetsa tse tharo." E thehiloe ka 2006, e se e holile ka potlako ho ba sebatli sa pele sa dijithale sa Europe mme e fana ka phihlello ea lits'ebeletso tsa dijithale tse 300 lefatšeng ka bophara. Ha e le hantle, Rebeat e tsamaisa mekhoa ea ho boloka libuka le ho sebetsana le mosebetsi oa morao-rao, joalo ka ho bapisa likarolo tsa data ho software ea accounting, kahoo lefapha la likarabello le lokolohile ho laola lichelete. Ba boetse ba fana ka lits'ebetso tsa ho tsamaisa litefo tsa borena ho latela litumellano tsa konteraka, litumellano tse tobileng le mabenkele a 'mino oa dijithale, ho hlahisa li-graph ho latela thekiso, mme ho bohlokoa ka ho fetisisa, ho romella data ho lifaele tsa CSV.

Ehlile, tšebeletso e tla le theko. Forbes e tlalehile hore mangolo a rekotiloeng a tlameha ho sebelisa Rebeat e le morekisi e le hore a ka fumana data ea k'hamphani, e bitsang khomishene ea thekiso ea 15% le tefiso e tsitsitseng ea $ 649 selemo se seng le se seng. Leha ho le joalo, likhakanyo li bontša hore maemong a mangata, tlaleho ea leibole e koahelang chelete e ngata haholo, ho bolelang hore ho saena le Rebeat ho ka pholosa chelete.