Ingabe udinga ukukhipha ngokushesha nangokuzenzakalelayo ulwazi kumadokhumenti amaningi? Futhi ingabe zigcinwa ngendlela yokuskena noma izithombe? Unenhlanhla uma uyikhasimende le-Amazon Web Services (AWS). I-Amazon imemezele ukuvulwa kokufinyelela ku , isevisi esekelwe emafini futhi ephethwe ngokugcwele esebenzisa umshini wokufunda ukuze ihlaziye amathebula, amafomu ombhalo, nawo wonke amakhasi ombhalo ngamafomethi we-elekthronikhi adumile. Okwamanje, izotholakala kuphela ezifundeni ezikhethiwe ze-AWS, ikakhulukazi eMpumalanga ye-US (e-Ohio naseNyakatho Virginia), eNtshonalanga US (Oregon), nase-EU (Ireland), kodwa ngonyaka ozayo i-Textract izotholakala kuwo wonke umuntu.
Ngokusho kwe-Amazon, i-Textract isebenza kahle kakhulu kunezinhlelo ezijwayelekile zokuqaphela izinhlamvu. Kusuka kumafayela agcinwe ebhakedeni le-Amazon S3, ingakwazi ukukhipha okuqukethwe kwezinkambu namathebula ngokusekelwe kumongo lapho lolo lwazi lwethulwa khona, njengokugqamisa ngokuzenzakalelayo amagama nezinombolo Zokuvikeleka Komphakathi kumafomu entela noma isamba samarisidi athathwe izithombe. Njengoba i-Amazon iphawula ku , I-Textract isekela amafomethi wezithombe ezifana nokuskena, ama-PDF nezithombe, futhi isebenza ngempumelelo nomongo kumadokhumenti aqondene nezinsizakalo zezezimali, umshwalense nokunakekelwa kwezempilo.
Imiphumela yezitolo ze-Textract ngefomethi ye-JSON, echazwe ngezinombolo zekhasi, izigaba, amalebula efomu, nezinhlobo zedatha, futhi ngokuzithandela ihlanganisa nesizindalwazi nezinsizakalo zokuhlaziya ezifana ne-Amazon Elasticsearch Service, i-Amazon DynamoDB, i-Amazon Athena, nemikhiqizo yokufunda yomshini. njenge-Amazon Comprehend , I-Amazon Comprehend Medical, i-Amazon Translate, ne-Amazon SageMaker ukuze icutshungulwe. Kungenjalo, idatha ekhishiwe ingadluliselwa ngokuqondile kumasevisi amafu ezinkampani zangaphandle ngezinjongo zokubalwa kwezimali nokuhlolwa kocwaningo noma ukusekela ukusesha okuhlakaniphile kwezingobo zomlando zamadokhumenti. Ngokusho kwe-Amazon, i-Textract ingakwazi “ngokunembile” ukucubungula izigidi zamakhasi emibhalo ehlukene “ngamahora ambalwa nje.”
Amakhasimende amaningi e-AWS asevele asebenzisa i-Textract, okuhlanganisa i-Globe and Mail, Isevisi Yesimo Sezulu Kazwelonke yase-UK, i-PricewaterhouseCoopers, inhlangano yokunakekela ephethwe engenzi nzuzo i-Healthfirst, nezinkampani ezishintshayo zenqubo yamarobhothi i-UiPath, Ripcord, neBlue Prism. I-Candor, isiqalo esihlose ukuletha obala embonini yezindlu ezibolekwayo, isebenzisa i-Textract ukuze ikhiphe idatha kumadokhumenti afana nezitatimende zasebhange, ama-stubs okukhokha kanye nemibhalo yentela ehlukahlukene ukuze kusheshiswe inqubo yokugunyaza imalimboleko kumakhasimende ayo.
"Amandla e-Amazon Textract ukuthi ikhipha ngokunembile umbhalo nedatha ehlelekile kunoma iyiphi idokhumenti ngaphandle kwesidingo sokufunda ngomshini okuthuthukile," kusho uSwami Sivasubramanian, iphini likamongameli we-Amazon Machine Learning. "Ngaphezu kokuhlanganiswa nezinye izinsizakalo ze-AWS, umphakathi omkhulu okhula eduze ne-Amazon Texttract uvumela amakhasimende ethu ukuthi athole inani langempela kumaqoqo awo amafayela, asebenze ngokuphumelelayo, athuthukise ukuthobelana nokuphepha, afake idatha ngokuzenzakalelayo, futhi asheshise izinqumo zebhizinisi."
Ngezansi ungabuka isethulo se-Textract enkomfeni ye-re:Invent 2018 ngesiNgisi.
Source: 3dnews.ru