Ukucutshungulwa kwe-skali okuqondakalayo kwe-stacing ehlakaniphile usebenzisa i-Amazon Bedrock Data automation
Ukucutshungulwa kwedokhumenti ye-Intelligent (IDP) ubuchwepheshe bokushintsha ukukhishwa, ukuhlaziya, kanye nokutolika imininingwane ebucayi kusuka ezinhlobonhlobo zemibhalo. Ngokusebenzisa i-Advanced Machine Learning (ML) kanye nama-algorithms wemvelo wokucubungula ama-algorithms, izixazululo ze-IDP zingakhipha kahle futhi zicubungule idatha ehlelekile embhalweni ongahleliwe, ukuhanjiswa komsebenzi we-cell-cebhuling.
Lapho kuthuthukiswa ngamakhono we-AI akhiqizayo, i-IDP inika amandla izinhlangano ukuguqula ukugeleza komsebenzi kwedokhumenti ngokusebenzisa ukuqonda okuthuthukile, ukukhishwa kwedatha okuhlelekile, nokuhlukaniswa okuzenzakalelayo. Izisombululo ze-IDP ezikhiqizwayo ze-AI-Powerment zingaphatha kangcono izinhlobonhlobo zamadokhumenti amamodeli wendabuko wendabuko kungenzeka angakuboni ngaphambili. Le nhlanganisela yezobuchwepheshe inomthelela kuzo zonke izimboni eziningi, kufaka phakathi izinsizakalo zokusekelwa kwezingane, umshuwalense, ukunakekelwa kwempilo, izinsizakalo zezezimali kanye nomkhakha womphakathi. Ukucutshungulwa kwencwajana yendabuko kwakha amabhodlela ama-bottonlenecks futhi kwandisa ingozi yephutha, kepha ngokusebenzisa lezi zixazululo ezithuthukile, izinhlangano zingathuthukisa ngokumangazayo ukusebenza kahle komsebenzi wazo kanye namakhono okubuyisa imininingwane. Izisombululo ze-IDP ezithuthukisiwe ze-AI zithuthukisa ukulethwa kwezidingo ngenkathi kunciphisa umthwalo wokuphatha kuzo zonke izimo ezahlukahlukene zokucubungula imibhalo.
Le ndlela yokucubungula idokhumenti ihlinzeka ngokusetshenzelwa okulinganiselwe, okusebenzayo, kanye ne-High-value Makhodem Descring eholela ekukhiqizweni okuthuthukile, izindleko ezincishisiwe, kanye nokwenza izinqumo ezithuthukisiwe. Amabhizinisi amukela amandla we-IDP augmented nge-generative ai angazuza ngokusebenza kahle, okuhlangenwe nakho kwamakhasimende okuthuthukisiwe, nokukhula okusheshayo.
Ku-blog post sciperable i-scallent Imebhu yedokhumenti ehlakaniphile usebenzisa i-Amazon Bedrock, sikhombise ukuthi singayakha kanjani ipayipi le-IDP elilinganiselwe usebenzisa ama-anthropic Foundation model e-Amazon Bedrock. Yize leyo ndlela ihanjiswa ukusebenza okuqinile, ukwethulwa kwe-Amazon Bedrock Data automation kuletha izinga elisha lokusebenza kahle nokuguquguquka kwezixazululo ze-IDP. Lokhu okuthunyelwe kuhlola ukuthi ezenzakalelayo ze-Amazon Bedrock Zembano zithuthukisa kanjani amakhono okucubungula amadokhumenti futhi kuqondiswe uhambo oluzenzakalelayo.
Izinzuzo ze-Amazon Bedrock Data automation
I-Amazon Bedrock Data Automation yethula izici eziningana ezithuthukisa kakhulu ukuqina nokunemba kwezixazululo ze-IDP:
- Izikolo zokuzethemba kanye nemininingwane yebhokisi lokubopha – I-Amazon Bedrock Data automation automation inikeza izikolo zokuqiniseka kanye nedatha yebhokisi lokubopha, ukuthuthukisa ukucacisa kwedatha nokusobala. Ngalezi zici, ungahlola ukuthembeka kwemininingwane ekhishwe, okuholele ekwenziweni kwezinqumo okwengeziwe. Isibonelo, izikolo eziphansi zethemba zingasayina isidingo sokubuyekezwa okwengeziwe komuntu noma ukuqinisekiswa kwamasimu athile wedatha.
- Imibhalo emifushane yokuthuthuka okusheshayo – I-Amazon Bedrock Data automation automation inikezela ngemibhalo eyakhelwe ngaphambili eyakha i-Creation of Desigred of Desicud Pippelines, kukusiza ukuthuthukisa kanye nokuhambisa izixazululo ngokushesha. I-Amazon Bedrock Data automation automation inikeza ukucushwa okuguqukayo kokukhipha ukuze kuhlangatshezwane nezidingo ezahlukahlukene zemibhalo ecubungula. Ukuze uthole amacala alula okusetshenziswa (i-OCR ne-Layout) noma ukukhishwa okulinganiselwe kombhalo kumadokhumenti, ungasebenzisa umphumela ojwayelekile. Ngokwemiphumela eyenziwe ngokwezifiso, ungaqala kusuka ekuqaleni ukuklama i-schema eyingqayizivele yokukhipha, noma usebenzise ama-blueprints anqabile kusuka kukhathalogi yethu njengendawo yokuqala. Ungenza ngokwezifiso i-Bluepprint yakho ngokuya ngezinhlobo zakho zemibhalo ethile nezidingo zebhizinisi zokubuyiselwa kolwazi okuhlosiwe futhi okunembile.
- Ukuxhaswa kokuhlukaniswa okuzenzakalelayo – I-Amazon Bedrock Data automation automation ye-automation kanye namadokhumenti afanayo nemibhalo enemibala efanelekile, okuholela ekuhlukaniseni amadokhumenti aqondile. Lesi sifiso esihlakaniphile sidambisa isidingo sokuhlungwa kwedokhumenti yezandla, ukunciphisa kakhulu ukungenelela komuntu kanye nokusheshisa isikhathi sokucubungula.
- Ukujwayelekile – I-Amazon Bedrock Data automation automation a automation automation ajwayelekile ngohlaka lwayo olujwayelekile, olusebenzisa zombili izinkambu ezijwayelekile (imephu amalebula amahle enkambu kumafomethi ajwayelekile) kanye nenani ledatha (ukuguqula idatha ekhishwe kwimininingwane). Le ndlela ejwayelekile yokwenziwa okujwayelekile isiza ukunciphisa izinhlangano zokucubungula idatha, ngakho-ke izinhlangano zingaguqula ngokuzenzakalela ukukhishwa kwedokhumenti eluhlaza kube yidatha esezingeni elilinganiselayo ehlanganisa kahle nezinhlelo zazo ezikhona kanye nokuhamba komsebenzi.
- Ukuguqulwa – Isici sokuguqulwa kwe-automation sedrock se-automation automation automation aguqukayo kuguqula izinkundla eziyinkimbinkimbi zedokhumenti zibe yidatha ehlelekile, elungele ibhizinisi ngokuzenzakalela ukuhlukanisa imininingwane ehlanganisiwe (njengamagama) ezingxenyeni ezingekho emthethweni. Lokhu kunamandla okwenziwa lula ukuthi izinhlangano zisingatha kanjani amafomethi wedokhumenti ahlukahlukene, ukusiza amaqembu achaze izinhlobo zezinkambiso zedatha nobudlelwano bensimu ehambelana nezinhlelo zabo zedatha ezikhona.
- Ukuqinisekisa – I-Amazon Bedrock Data automation ithuthukisa ukunemba kokusebenza kwedokhumenti ngokusebenzisa imithetho yokuqinisekisa ezenzakalelayo yedatha ekhishwe, amabanga ezinombolo, amafomethi amakhethini, amasheke wenkambu. Lolu hlelo lokuqinisekisa lusiza izinhlangano ukuthi zikhombe izindaba zekhwalithi yedatha, izibuyekezo zabantu lapho zidingeka, futhi ziqiniseke isiqiniseko semininingwane ekhishwe ihlangabezana nemithetho ethile yebhizinisi nezidingo zokuhambisana ngaphambi kokungena ezinhlelweni ezansi nomthetho ngaphambi kokungena ezinhlelweni ezansi nomthetho ngaphambi kokungena ezinhlelweni ezansi nomthetho ngaphambi kokungena ezinhlelweni ezansi nomthetho ngaphambi kokungena ezinhlelweni ezansi nomthetho
Ukubuka konke
Umdwebo olandelayo ukhombisa ukwakheka okuphelele kusengaphambili okusebenzisa i-Amazon Bedrock Data automation kanye nemisebenzi yezinyathelo ze-AWS kanye ne-Amazon Agegmented AI (Amazon A2i) ukuhlinzeka ngezindleko ezisebenzayo zokuthola imithwalo yedokhumenti.
Imisebenzi yesinyathelo ukusebenza icubungula izinhlobo eziningi zemibhalo kubandakanya ama-pdfs amaningi nezithombe usebenzisa i-Amazon Bedrock Data automation. Isebenzisa ama-blueprints we-automation we-automation we-automation we-automation e-automation e-automation e-Amazon ahlukahlukene
Ukuhamba komsebenzi kusebenza ifayela (PDF, JPG, PNG, TIFF, Doc, Docx) equkethe idokhumenti elilodwa noma amadokhumenti amaningi ngezinyathelo ezilandelayo:
- Ngemibhalo enamakhasi amaningi, ukuqhekeka kanye nemingcele yedokhumenti enengqondo
- Qondanisa idokhumenti ngalinye kwi-blueprint efanelekile
- Kusebenza imiyalo ethile ye-extraction ekhethekile ye-bluepprint ukuthola imininingwane kusuka kudokhumenti ngalinye
- Yenza ukujwayelekile, ukuguqulwa kanye nokuqinisekiswa kwedatha ekhishwe ngokuya ngomyalo ochazwe nge-blueprint
Isinyathelo semisebenzi yemephu sisetshenziselwa ukucubungula idokhumenti ngalinye. Uma idokhumenti ihlangabezana nomkhawulo wokuzethemba, umphumela uthunyelwa kwi-Amazon Simple Service Service (Amazon S3) Ibhakede. Uma noma yimiphi imininingwane ekhishwe iwela ngaphansi komkhawulo wokuzethemba, lo mbhalo uthunyelwa ku-Amazon A2i ukuze ubukeke. Ababuyekezi basebenzisa i-Amazon A2I UI ngebhokisi lokubopha liqokomisa amasimu akhethiwe ukuqinisekisa imiphumela yokukhipha. Lapho ukubuyekezwa komuntu kuqediwe, ithokheni le-Callback Task lisetshenziselwa ukuqala kabusha umshini wombuso kanye nokuphuma kobuntu kuthunyelwa kubhakede le-S3.
Ukuthumela lesi sixazululo ku-akhawunti ye-AWS, landela izinyathelo ezinikezwe endaweni yokugcina ehambisana nalesi siga.
Ezingxenyeni ezilandelayo, sibuyekeza i-automation ye-automation ye-automation ethile ye-Amazon Bedrock eyenziwe yathunyelwa kusetshenziswa lesi sixazululo, sisebenzisa isibonelo sefomu lokubhalisa kwengane.
Ukuhlukaniswa okuzenzakalelayo
Ekusetshenzisweni kwethu, sichaza igama lesigaba sedokhumenti nge-bleatherprint ngayinye yezenkantolo, njengoba kuboniswe esikrinini esilandelayo. Lapho kucubungula izinhlobo eziningi zemibhalo, njengamalayisense okushayela kanye namafomu okubhalisa izingane ngokuzenzakalela, uhlelo lusebenzisa ngokuzenzakalelayo i-blueprint efanelekile ngokususelwa ekuhlaziyweni kokuqukethwe, ukuqiniseka ukuthi i-logic yokukhishwa isetshenziselwa uhlobo ngalunye lwedokhumenti.
Idatha ejwayelekile
Sisebenzisa idatha ejwayelekile ukwenza isiqiniseko sokuqiniseka ukuthi amasistimu aphansi athola idatha efomethwe ngokufanayo. Sisebenzisa zombili lezi zinto ezikhishwe ngokusobala (ngolwazi olucacisiwe olubonakalayo kudokhumenti) nokukhishwa okuphelele (ngolwazi oludinga uguquko). Isibonelo, njengoba kukhonjisiwe ku-skrini elandelayo, izinsuku zokuzalwa zilinganiswa ngefomethi ye-YYYY-MM-DD.
Ngokufanayo, ifomethi yezinombolo zokuphepha komphakathi iguqulwa ibe yi-XXX-XX-XXXX.
Ukuguqulwa Kwedatha
Ngohlelo lokubhaliswa kwengane, sisebenzise ukuguqulwa kwedatha ngokwezifiso ukuvumelanisa idatha ekhishwe ngezidingo ezithile. Isibonelo esisodwa uhlobo lwemininingwane yethu yangokwezifiso yamakheli, aphula amakheli amugqa owodwa abe yizinkambu ezihlelekile (umgwaqo, idolobha, isimo, i-zipcode). These structured fields are reused across different address fields in the enrollment form (employer address, home address, other parent address), resulting in consistent formatting and straightforward integration with existing systems.
Ukuqinisekiswa kwedatha
Ukuqaliswa kwethu kufaka imithetho yokuqinisekisa yokugcina ukunemba kwedatha nokuhambisana. Ngokwesibonelo sethu sokusebenzisa, sisebenzise ukuqinisekiswa okubili: 1. Qinisekisa ukuba khona kwesiginesha ka-ekerolae kanye no-2. Qinisekisa ukuthi usuku olusayiniwe alukho esikhathini esizayo.
Isikrini esilandelayo sibonisa umphumela wemithetho engenhla yokuqinisekiswa esetshenziswe kudokhumenti.
Ukuqinisekiswa komuntu-in loop
Isithwebuli esilandelayo sibonisa inqubo yokukhishwa, okubandakanya amaphuzu okuqiniseka futhi kuhlanganiswe nenqubo yoku-Jehova-in loop. Kubonisa nokujwayelekile okusetshenziselwa usuku lokuzalwa.
Ukugcina
I-Amazon Bedrock Data automation automation athuthuka i-IDP ngokungenisa amaphuzu ukuzethemba, idatha yebhokisi lokubopha, ukuhlukaniswa okuzenzakalelayo, kanye nokuthuthukiswa okusheshayo ngokusebenzisa ama-blueprints. Kulokhu okuthunyelwe, sikhombise ukuthi singasebenzisa kanjani ithuba lokuthi amakhono alo athuthukile okujwayelekile kwedatha, inguquko, kanye nokuqinisekiswa. Ngokuthuthukela kwi-Amazon Bedrock Data automation automation, izinhlangano zinganciphisa kakhulu isikhathi sokuthuthuka, zithuthukise ikhwalithi yedatha, bese zidala izixazululo ze-IDP eziqinile, ezinamandla ze-IDP ezihlanganisa nezinqubo zokubuyekezwa kwabantu.
Landela ibhulogi lokufunda lomshini we-AWS ukugcina usesikhathini ngamakhono amasha futhi usebenzise amacala e-Amazon Bedrock.
Mayelana nababhali
Abdul Navaz Ingabe ukwakhiwa kwezixazululo eziphezulu e-Amazon Web Services (AWS) Team Team Team Teams) Team, kususelwa eDallas, eTexas. Njengoba iminyaka engaphezu kwe-10 yesipiliyoni ema-arts, ugxile ezixazululweni zesimanje zokusekelwa kwezingane kanye nezinhlaka zezenhlalakahle zezingane usebenzisa izinsizakalo ze-AWS. Ngaphambi kokuba yiqhaza lakhe njengomakhi wohlelo lwezinhlelo zokusebenza, ama-navaz asebenze njengonjiniyela ophezulu wokuxhaswa kwamafu, enakekela izixazululo zokuxhumana.
Venkata kambana Ingabe ukwakhiwa kwezixazululo okuphezulu eqenjini lezempilo le-Amazon Web Services (AWS) lisuselwa eSacramento, eCalif. Kule ndima, usiza amakhasimende omkhakha womphakathi afezekisa izinhloso zawo zemishini ezingokwemishini.
Sanjeev gelelapaka kungumakhi wezakhiwo eziphambili ze-Solution kanye ne-AI ukuhola komkhakha womphakathi. USanjeev ungumbhali oshicilelwe ngamabhulogi ambalwa kanye nencwadi ekhiqizwayo AI. Futhi uyisikhulumi esaziwayo emicimbini eminingana efaka i-RE: Ukusungula kanye neNgqungquthela. USanjeev uneziqu zeziqu zobunjiniyela ezivela endabeni ye-Indian Institute of Technology kanye ne-MBA evela e-University of Notre Dame.
