Brainware Automated Intelligent Capture Drives High Value Experiences
Brainware Automated Intelligent Capture Drives High. Value Experiences Across Campus David Luzier – Vice President, Engineering
• Transcript Challenges • Brainware Intelligent Capture Agenda – Solutions • Brainware for Transcripts – Tables Advanced to Complex • Robotic Process Automation
Transcript Challenges Too Much Info…Too Little Time
What does Transcript Processing look like today?
Challenges with Transcripts: • 3. 6 million high school students graduated in 2018 – and nearly 70% of them will enroll in college for 2018 -19 • More than 1/3 of college students have transferred at least once • 40% of transfer students have attended 3 or more colleges or universities • The average college transcript is 3 pages long • Paper is still the most common transcript format with PDF increasing in popularity • Schools have not fully utilized XML/EDI
Brainware Intelligent Capture
Input Capture paper, email, fax, and electronic document OCR Classify Extract Sample content slide Read data on the document Classify the type of document (invoice, check request, statement, correspondence, etc. ) Extract header and line item data Validate data against other external sources Exception management Reporting, business intelligence, process insight Integrate Independent of target system
Brainware for Transcripts Transcript engine extracts data from complex tables Extract, validate and export key data: Applicant, Institution, Course/Grade Faster Processing = Admissions can review and select qualified candidates faster
College High school Military Classify • Context-based • Document agnostic • Pre-loaded with transcript sample set
Field Extraction • Learned context around data you need • No need to explicitly define templates, anchors, zones or specific keywords • Just a few transcript examples enable processing of many different types and formats • Pre-loaded transcript sample set 68% 88% 70% 71%
Invoice Table
Transcript Table
Stacked Transcript Table
Are you Kidding? ? ? ? ? ?
Advanced to Complex – – – – “Stacked” Tables Search Approach Repeated Patterns Leverage Headers Leverage Formats Multi-row Labels Column Formats
01010101 01010101 Reconcile Using External Data • Fault-tolerant search reconciles extracted data against known source • Independent of misspellings, abbreviations and OCR errors • Increases accuracy and promotes pass-through processing
Robotic Process Automation
Robotic Process Automation • In 2017 Forrester projected that the RPA market would be worth $2. 9 billion by 2021 – $250 M in 2016 • The value prop of RPA is similar to that of Intelligent Data Capture – Increase efficiencies – Decrease manual labor costs
Robotic Process Automation The major difference is that typical processes addressed by RPA have traditionally been repeatable and included little variance. This has enabled them to be fairly hard-coded and to not have to deal with a lot of exceptions. In contrast, document capture is more specialized and deals with a significant number of exceptions. But, end users don’t typically understand the differences, they only see the similarities. As they receive budget for large RPA implementations, they eventually run into document processes that they expect their RPA solutions to automate. Only, RPA can’t do it, at least [not] without an assist from document capture. And increasingly, we are seeing more and more of that. Document Imaging Report April 20, 2018
Robotic Process Automation
Machine Learning and RPA
Machine Learning and RPA • Combination of Advanced Machine Learning and RPA • Dynamic Variance Network – Learning • Input – Field Text with or without positional information • Calculate vectors between all OCR words on the document and the target field – Extraction • Overcome OCR errors • As field shifts many vectors shift as well
Robotic Process Automation
Conclusion • AI and Advanced Machine Learning – Increase efficiencies – Decrease manual labor costs • RPA Combined with Machine Learning – Dynamic Variance Network (DVN) • • Continues to improve efficiency gains Takes guess work of what a user trains or not Addresses shortcomings of deviations with RPA Addresses shortcomings “definitions”, “thumbprint”, “templates”
Brainware for Invoices Multi-client capability (SSC, BPO) Global Support – 27 languages Extracts, validates and exports transaction data to ERP or ECM Extracts international tax types (HST, PST, GST, ICMS, VAT) Automation = fewer FTEs required Faster processing = more supplier discounts
Turn Documents into Data with Automated Intelligent Capture • Wednesday 9: 10 – 9: 20 am and 5: 30 – 5: 40 pm • Thursday 12: 10 – 12: 20 pm and 3: 50 – 4: 00 pm Stop Managing Your Data & Start Using it with Content Services • Wednesday 12: 30 – 12: 40 pm and 4: 50 – 5: 00 pm • Thursday 12: 30 – 12: 40 pm and 3: 30 – 3: 40 pm Make Your Content Work for You with Eforms & Workflow • Wednesday 1: 10 – 1: 20 pm and 5: 10 – 5: 20 pm • Thursday 9: 10 – 9: 20 am and 12: 50 – 1: 00 pm Keep Your Content Secure & Compliant • Wednesday 12: 50 – 1: 00 pm and 4: 30 – 4: 40 pm • Thursday 9: 25 – 9: 35 am and 1: 10 – 1: 20 pm
? ? Questions ?
Thank you David Luzier Vice President, Engineering David. Luzier@Hyland. com
- Slides: 29