Hlola amadokhumenti anamakhasi amaningi ngokubuyekezwa komuntu usebenzisa i-Amazon Bedrock Data automation ne-Amazon Sagemaker AI

Izinhlangano ezizophesheya kwezimboni zibhekana nezinselelo ezinamavolumu aphezulu amadokhumenti amaningi adinga ukucubungula okukhaliphile ukuze kukhishwe imininingwane efanele. Yize i-automation sekenze ngcono le nqubo, ubuchwepheshe bomuntu busadingeka ezimweni ezithile ukuqinisekisa ukunemba kwedatha nekhwalithi.

Ngo-Mashi 2025, ama-AWS wethula ama-automation wedatha ye-Amazon Bedrock, enika amandla abathuthukisi ukuthi asebenzise isizukulwane sokuqonda okubalulekile kokuqukethwe okuyisisekelo kwe-multimodal, kufaka phakathi amadokhumenti, amadokhumenti, ividiyo, kanye nomsindo. I-Amazon Bedrock Data automation autorylines ukuguqulwa kwedokhumenti yokusebenza kokusebenza ngokuzenzakalela ukukhishwa, ukuguqulwa, kanye nesizukulwane sokuqonda okuqukethwe okungahleliwe. Inciphisa imisebenzi edla isikhathi efana nokulungiselela idatha, ukuphathwa kwamamodeli, ukulungiswa okuhle, ubunjiniyela obusheshayo, kanye nokufaka ama-api-uku-API okuhlangene, nokuthumela izintengo ezimbalwa ngezindleko eziphansi kunezindleko eziphansi kunezindleko eziphansi kunezinkinga eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezinkinga eziphansi kunezinkinga eziphansi kunezinkinga eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezinhlobo eziphansi kunezindlela eziphansi.

I-Amazon Bedrock Data automation yenza lula imisebenzi eyinkimbinkimbi yokwenza amadokhumenti, kufaka phakathi ukuhlukaniswa kwamadokhumenti, ukuhlukaniswa, ukukhishwa, ukuqinisekiswa, ngenkathi kufakwa ukubonwa okubonakalayo ngezinto zokuzimela, ukuhlinzeka ngemininingwane ethembekile emithonjeni yedatha engahleliwe. Kodwa-ke, yize amakhono athuthukile we-Amazon Bedrock Detation automation athumela ezishintshayo ezihlukile, kusele lapho kuhlala khona lapho ukwahlulela kwabantu kubaluleke kakhulu. Yilapho ukuhlanganiswa ne-Amazon Sagemaker AI kwakha isixazululo esinamandla sokugcina. Ngokufaka ama-loops wokubuyekezwa komuntu ekusebenzeni komsebenzi wokusebenza kwedokhumenti, izinhlangano zingagcina amazinga aphezulu okunemba ngenkathi egcina ukusebenza kahle. Nge-Loop Yokubukeza Yomuntu, izinhlangano zingaba:

• Qinisekisa izibikezelo ze-AI lapho ukuzethemba kuphansi
• Phatha amacala asenqenqemeni nokudelela ngempumelelo
• Gcina ukuhambisana okulawulwayo ngokubhekisisa okufanele
• Gcina ukunemba okuphezulu ngenkathi kukhulisa ezenzakalelayo
• Dala ama-loops okuphendula ukuze uthuthukise ukusebenza kwemodeli ngokuhamba kwesikhathi

Ngokusebenzisa izihibe zabantu ngokunemba, izinhlangano zingagxila ekunakekelweni kwabantu ezingxenyeni ezingaqinisekile zamadokhumenti ngenkathi zivumela izinhlelo ezenzakalelayo ukuphatha ukukhishwa okujwayelekile, ukudala ibhalansi efanelekile phakathi kokusebenza kahle nokunemba. Kulokhu okuthunyelwe, sikhombisa ukuthi ukucubungula kanjani amadokhumenti amakhasi amaningi nge-loop yokubukeza komuntu usebenzisa i-automation ye-Amazon Bedrock ne-sagemaker ai.

Ukuqonda izikolo zokuzethemba

Izikolo zokuzethemba zibalulekile ekunqumeni ukuthi zibuyisele nini ukubuyekezwa kwabantu. Izikolo zokuzethemba zingamaphesenti wokuqiniseka okuzenzakalelayo kwedatha ye-Amazon Bedrock okuthi ukukhishwa kwakho kunembile.

Umgomo wethu ukwenza lula ukucubungula kwedokhumenti okukhaliphile (i-IDP) ngokuphatha ukuphakanyiswa okusindayo kokubalwa kokunemba ngaphakathi kwe-Amazon Bedrock Data automation. Lokhu kusiza amakhasimende agxile ekuxazululeni izinselelo zebhizinisi labo nge-Amazon Bedrock Data automation esikhundleni sokubhekana nezinqubo eziyinkimbinkimbi zokushaya amagoli. I-Amazon Bedrock Data automation yenza amamodeli ayo ngephutha lokulinganisa elindelekile (ECE), i-metric esiza ukulinganisa okungcono, okuholela ezikolo ezethembekile futhi ezinembile zokuzethemba.

Ekusebenzeni kokusebenza kwedokhumenti, izikolo zokuzethemba zivame ukuhunyushwa ngokuthi:

• Ukuzethemba okuphezulu (90-100%) – Ukuqiniseka okuphezulu ngokukhishwa kwayo
• Ukuzethemba okuphakathi (70-89%) – Ukuqiniseka okunengqondo okunamandla athile wephutha
• Ukuzethemba okuphansi (<70%) – Ukungaqiniseki okuphezulu, okungenzeka kudinga ukuqinisekiswa komuntu

Sincoma ukuvivinya i-Amazon Bedrock Data ezenzakalelayo emininingwaneni yakho ethile ukunquma umkhawulo wokuzethemba odala ukuvela kokusebenza komuntu.

Ukubuka konke

Ukwakhiwa okulandelayo kuhlinzeka ngesisombululo esingenasici sokucubungula amadokhumenti ahlukahlukene ngama-loops wokubuyekezwa komuntu usebenzisa i-automation ye-Amazon Bedrock ne-sagemaker ai.

Ukuchitheka komsebenzi kuqukethe lezi zinyathelo ezilandelayo:

1. Amadokhumenti alayishwa kwisevisi yokugcina elula ye-Amazon (i-Amazon S3) Ibhakede lokufaka, elisebenza njengephuzu lokungena lemibhalo elicutshungulwe nge-Amazon Bedrock Data automation.
2. Umthetho we-Amazon Wearnbridge ubona ngokuzenzakalelayo izinto ezintsha kubhakede le-S3 futhi asuse imisebenzi yezinyathelo ze-AWS zisebenza ngokusebenza okusebenza ama-PIPEline elokucubungula idokhumenti.
3. Ngaphakathi kwemisebenzi yesinyathelo isebenza ngokusebenza komsebenzi, The bda-document-processor Umsebenzi we-AWS Lambda uyabulawa, okhipha i-Amazon Bedrock Detation automation nge-blueprint efanelekile. I-Amazon Bedrock Data automation isebenzisa le miyalo ebekiwe yokukhipha nokucubungula imininingwane evela kudokhumenti.
4. I-Amazon Bedrock Data automation ihlaziya idokhumenti, ikhipha amasimu asemqoka ngezikolo ezenzelwe ukuzithoba, futhi igcina umphumela ocutshunguliwe kwenye i-S3 Bucket. Lokhu kukhipha kuqukethe imininingwane ekhishwe namazinga okuzethemba ahambisanayo.
5. Imisebenzi yesinyathelo isebenza ngokusebenza kunxusa bda-classifier Umsebenzi weLambda, othatha ubuyisa umbhede we-Amazon Bedrock automation automation automation automation kusuka e-Amazon S3. Lo msebenzi uhlola ukuzethemba izikolo ngokumelene nemikhawulo echaziwe yamasimu akhishwe.
6. Okwezinkambu ezinezikolo zokuzethemba ngaphansi komkhawulo, ukugeleza kwezinyawo zomsebenzi lapho idokhumenti idokhumenti eSagemaker AI yokubuyekezwa komuntu. Usebenzisa i-UI yangokwezifiso, abantu babuyekeza imisebenzi futhi baqinisekise izinkambu ezivela kumakhasi. Ababuyekezi bangalungisa izinkambu ezikhishwe ngokungafanele yinqubo ezenzakalelayo.
7. Idatha yefomu eliqinisekisiwe nelilungisiwe elivela ekubuyekezweni komuntu ligcinwa kubhakede le-S3.
8. Lapho umphumela we-sagemaker ai ubhaliwe e-Amazon S3, ukhiphe bda-a2i-aggregator Aws Lambda evuselela i-Payload ye-Amazon Bedrock Data automation automation automation automation ngenani elisha elibukezwe ngumuntu. Lo mphumela ohlanganisiwe ugcinwa e-Amazon S3. Lokhu kuhlinzeka ngokugcina kokugcina, okuphezulu okuqiniseka okulungele izinhlelo eziphansi.

Izimfuneko

Ukuze usebenzise lesi sixazululo, udinga i-AWS Cloud Development Kit (AWS CDK), i-Node.js, neDocker efakwe emshinini wakho wokuhanjiswa. Isikripthi sokwakha senza ukufakwa kanye nokuhanjiswa kwesixazululo.

Sebenzisa Isixazululo

Qedela lezi zinyathelo ezilandelayo ukuthunga ikhambi:

1. Clone indawo yokugcina izixazululo emshinini wakho wokuhanjiswa.
2. Zulazulela kwisikhombi sephrojekthi bese usebenzisa iskripthi sokwakha:

./build.sh

Ukuhanjiswa kwakha izinsiza ezilandelayo kwi-AWS Akhawunti yakho:

• Amabhakede amabili amasha we-S3: eyodwa yokulayisha kokuqala kwemibhalo neyodwa yokuphuma kwemibhalo
• Iphrojekthi ye-automation ye-automation ye-automation ye-Amazon Bedrock nemibhalo emihlanu esetshenziselwa ukucubungula idokhumenti yokuhlola
• Ichibi lomsebenzisi le-Amazon Cognito labasebenzi abazimele ukuthi iqiniso le-Amazon SageMaker lihlinzeka nge-sagemaker ai yedatha engaphansi kwesikolo sokuzethemba
• Imisebenzi emibili ye-lambda kanye nokusebenza komsebenzi wesinyathelo okusetshenziselwa ukucubungula amadokhumenti wokuhlola
• Izithombe ezimbili ze-Amazon Elastic Direguer Registry (Amazon ECR) Izithombe ezisetshenziselwe imisebenzi yeLambDA ukucubungula amadokhumenti wokuhlola

Faka isisebenzi esisha kubasebenzi abazimele

Ngemuva kokuthi kwakhiwa sekuqediwe, kufanele ungeze isisebenzi kubasebenzi abazimele eqinisweni lomhlaba we-sagemaker. Qedela lezi zinyathelo ezilandelayo:

1. E-sagemaker ai console, ngaphansi Iqiniso Lomhlabathi Kwiphaneli yokuhambisa, khetha Ukulebula Amandla Work Workbese ukhetha -Ngaziwa muntu ithebhu.

2. Ku Abasebenzi Ingxenye, khetha Mema abasebenzi abasha.

3. Ingomane Amakheli E-imeyilifaka amakheli e-imeyili wabasebenzi ofuna ukumema. Ngalesi sibonelo, sebenzisa i-imeyili oyitholayo.
4. Qoka Mema abasebenzi abasha.

Ngemuva kokuthi isisebenzi sengeziwe, bazothola i-imeyili ene-password yesikhashana. Le nqubo ingathatha imizuzu emi-5 ngaphambi kokuba i-imeyili itholwe.

5. Use Ukulebula Amandla Work Work ikhasi, ku Isifinyezo sabasebenzi abazimele ingxenye, khetha isixhumanisi se Ilebuli i-URL yokungena ngemvume ye-portal.

6. Esikhathini esisheshayo, faka ikheli le-imeyili olisebenzise ngaphambili ukusetha isisebenzi futhi unikeze ngephasiwedi yesikhashana kusuka ku-imeyili, bese ukhetha Ngena ngemvume.

7. Nikeza iphasiwedi entsha lapho utshelwa.

Uzoqondiswa kabusha ekhasini lomsebenzi womugqa we-Work for abasebenzi abazimele. Esiqongweni sekhasi, isaziso sithi awulona ilungu leqembu lomsebenzi okwamanje. Kufanele ugcwalise lelo hlelo esinyathelweni esilandelayo ukuze uqiniseke ukuthi imisebenzi inikezwa kahle.

8. Use Ukulebula Amandla Work Work ikhasi, vula iqembu elizimele (kulokhu okuthunyelwe, bda-workforce).

9. Use Abasebenzi ithebhu, khetha Engeza izisebenzi eqenjini.

10. Faka isisebenzi esanda kuqinisekiswa eqenjini.

Hlola ikhambi

Ukuhlola ikhambi, layisha idokhumenti lokuhlola elisendaweni assets ifolda yephrojekthi kubhakede le-S3 elisetshenziselwe amadokhumenti angenayo. Ungabheka inqubekela phambili yohlelo emisebenzini ye-Step Console noma ngokubukeza izingodo nge-Amazon CloudWatch. Ngemuva kokuthi idokhumenti iyacutshungulwa, ungabona umsebenzi omusha ulayini womsebenzisi eSagemaker AI. Ukubuka lo msebenzi, zulahlela emuva ku Ukulebula Amandla Work Work ikhasi bese ukhetha isixhumanisi se Ilebuli i-URL yokungena ngemvume ye-portal.

Ngena ngemvume usebenzisa ikheli le-imeyili nephasiwedi ebuyekeziwe kusuka ekuqaleni. Uzobona ikhasi elibonisa imisebenzi okufanele ibuyekezwe. Khetha umsebenzi bese ukhetha Qala ukusebenza.

Kwi-UI, ungabuyekeza into ngayinye ebingaphansi kwesikolo sokuqiniseka (ifakwe kuma-70%) kudokhumenti elisetshenzisiwe.

Kuleli khasi, ungaguqula idatha kumanani alungisiwe. Idatha ebuyekeziwe izogcinwa kubhakede lokukhipha le-S3 ku a2i-output/bda-review-flow-definition//review-loop-/output.json Ifayela. Le datha ingahle icutshungulwe futhi isetshenziselwe ukuhlinzeka ngamanani alungisiwe ngolwazi olubuyiselwe kudokhumenti.

Hlanza

Ukuqeda yonke izinsiza ezenziwe kulesixazululo, gijima umyalo ogelezayo ovela kumkhombandlela wezimpande wephrojekthi

CDK Chitha

Ukugcina

Kulokhu okuthunyelwe, sikhombise ukuthi inhlanganisela ye-Amazon Bedrock Detomation automation kanye ne-sagemaker ai ihambisa ukusebenza kwezikhathi zokusebenza kanye nokunemba kwezinga labantu bobabili ukucubungula ikhasi elilodwa neliningi ledokhumenti ledokhumenti.

Sikukhuthaza ukuthi uhlole leli phethini ngezinselelo zakho zokwenza izinselelo. Isixazululo senzelwe ukuthi siguqulwe ngazo ezinhlotsheni ezahlukahlukene zemibhalo futhi senziwe ngezifiso ukufeza izidingo ezithile zebhizinisi. Zama ukuqaliswa okuphelele okutholakalayo kwi-GitHub Repository yethu, lapho uzothola khona yonke ikhodi nokucushwa okudingekayo ukuze uqalise.

Ukuze ufunde kabanzi ngedokhumenti Intelligence Solutions kuma-automation automation we-Amazon Bedrock kanye ne-SageMaker AI imibhalo.

Sicela wabelane ngolwazi lwakho kumazwana noma ufinyelele kubabhali ngemibuzo. Isakhiwo esijabulisayo!

Mayelana nababhali

UJoe Morotti Ingabe ukwakhiwa kwezixazululo e-Amazon Web Services (AWS), ukusebenza ngamakhasimende e-Financial Services e-US. Ubambe izinhlobo eziningi zezobuchwepheshe futhi ujabulele ukukhombisa ubuciko bekhasimende kungenzeka. Uyilungu elisebenzayo lemiphakathi yasensimini yezemidlalo ye-AWS yezobuchwepheshe ye-AI ne-Amazon Connect. Esikhathini sakhe samahhala, uyakujabulela ukuchitha isikhathi esisezingeni eliphakeme nomndeni wakhe ukuhlola izindawo ezintsha kanye nokuhlaziya ukusebenza kweqembu lakhe lezemidlalo.

I-Prashanth Ramanathan Ingabe ukwakhiwa kwezixazululo okuphezulu kuma-AWS, anothando mayelana nobuchwepheshe be-AI, obungenasisekelo kanye ne-database. Ungunjiniyela ophambili ophakeme kwi-Makhulu Emisebenzini Yezezimali futhi uhole ukuthutha kwamafu amakhulu kanye nemizamo yesimanje.

U-Andy Hall Ingabe ukwakhiwa kwezixazululo eziphezulu nge-AWS futhi kugxile ekusizeni amakhasimende we-Financial Services ngoguquko lwawo lwedijithali ku-AWS. U-Andy usize izinkampani ukwakhiwa, ukuthutha, kanye nezicelo zesilinganiso esikhulu kuma-AWS. Eminyakeni engama-30 edlule, u-Andy uhole imizamo ezungeze ukuthuthukiswa kwesoftware, ukwakhiwa kwesistimu, ukucubungula idatha, kanye nokuhamba komsebenzi wentuthuko kwamabhizinisi amakhulu.

UVikas shah Ingabe ukwakhiwa kwezixazululo e-Amazon Web Services esebenza ngokukhethekile kubuhlakani bemibhalo kanye nezixazululo ezinamandla ai. Umthandi wezobuchwepheshe, uhlanganisa ubuchwepheshe bakhe ekusebenzeni kwamadokhumenti, ukusesha okukhaliphile, kanye ne-ai ekhiqizayo ukusiza amabhizinisi asebenze kahle imisebenzi yawo. Indlela yakhe emisha yokuxazulula izinselelo zebhizinisi eziyinkimbinkimbi ezihambisana nokuphathwa kwamadokhumenti, ama-robotic, nobuchwepheshe obusafufusa.