Dynamicocr – People-augmented AI Solutions

DynamicOCR™

People-Augmented AI Solutions

The Business Problem

palm with papers

High-Value Business Data can be contained across many pages, many documents, and/or many pages deep within a scanned, faxed, or emailed document that have been received by your organization.  You can even have multiple loan documents, patient encounters, student enrollments, claims, etc. intermixed all within a single batch.  You’re dealing not only with a variety of document sources, but you are also having to contend with varying file formats, file sizes, trailing documents, an enormous volume of transactions and you are even having to contend with paper-based documents. 


Examples of High-Value Business Data include key data (used in a decision-making process) such as: loan numbers, member numbers, first and last names (including prefixes and suffixes), phone numbers, document names or types, prior authorization numbers, plan numbers, date of birth, etc.  To have a Knowledge Worker/Case Worker examine all these pages one by one, for the purposes of manually finding and extracting key, relevant High-Value Business Data is a very time consuming and has the high potential of transposing or entirely missing the key data all together.  Training your team of Knowledge Workers to manually Search, Identify and Extract the High-Value Business Data elements is also a very time-consuming process and becomes quite problematic when employee turnover occurs. 

Introducing DynamicOCR

DynamicOCR

DynamicOCR from CAPSYS Technologies is our most advanced OCR ExecuBOT technology that we have released to date. DynamicOCR performs the heavy-lifting behind the scenes by automatically targeting and extracting the High-Value Business Data that is needed downstream and/or elsewhere within your mission critical business processes and your line of business systems. High-Value Business Data that your Knowledge Workers today are painstakingly manually searching for - page by page, word by word – only to then manually re-key what the identified data into a document capture solution and/or elsewhere within an entirely different application or line of business system. 


DynamicOCR automatically detects and extracts text out of image and non-imaged based documents using a combination of proprietary and highly optimized Search | Identify | Extract technologies and mathematical algorithms. The Artificial Intelligence (AI) components of DynamicOCR refers to the ability of CAPSYS’ proprietary algorithms to select a value that yields the highest confidence level out of two or more possible results. The proprietary AI algorithms incorporate the use of three (3) unique and proprietary search techniques, in which we have defined as: Uninformed Search | Informed Search | Reverse Search technology. 
  • Uninformed Searches are performed when a document has been presented to the system and is not recognized as a known document.
  • Informed Searches are performed when a document has been presented to the system and is recognized as a known document.
  • Reverse Searches are performed when a document has been presented to the system in combination with one or more index values that are present within CAPSYS.


We have further augmented DynamicOCR’s capabilities by developing AI algorithms that allow it to recognize documents with previously seen layouts, which allows DynamicOCR to identify High-Value Business Data quicker and more accurately as the system processes more documents over time. Additionally, any corrections a user makes to DynamicOCR’s output will be processed and utilized with future documents. The combined usage of these various tools and Machine Learning techniques allows DynamicOCR to become smarter over a short period of time – building upon its own internal knowledgebase of High-Value Business Data that customers desire to capture, consume, and process.

No Templates Required

DynamicOCR completely jettisons any need to have predefined templates configured by a Systems Administrator in which the underlying OCR engine would traditionally use to target key areas of an incoming document so it may perform text extraction routines. As an alternative to the traditional approach of using pre-defined templates, DynamicOCR makes use of various complex scientific methods such as Artificial Intelligence (AI), Machine Learning (ML), Heuristics, and Mathematics to Search |Identify |Extract High-Value Business Data contained within your documents. 

Key Value Pairs

Keyvalues

Included within the DynamicOCR framework and architecture is a concept known as automatic Key Value Pair identification which dramatically accelerates identifying High-Value Business Data during the Search | Identify |Extract process. Key Value Pairs are defined as establishing a relationship between two related data items: a Key and a Value. The Key defines the data also known as an index and is fixed, and the Value is variable and describes the key. For example, in the case of an Invoice Document, the text contained on the physical document noted as “Invoice Number” will represent the Key and the actual invoice number “8754312” serves as the Value. 


Additionally, DynamicOCR provides two additional fields of relevant decision-making data besides Key and Value and those are described as: Type and Accuracy.


The Type data provides the nature of the content. For example: Account Number, Phone Number, Name, Member Number, etc.

The Accuracy data is a confidence level. This confidence level is computed by taking into account various parameters like OCR results at character and word levels, type of key, position on the page, and more.


These values can be used in CAPSYS’ workflow processes to determine what actions based on your business-rules need to be occur if these values fall outside of tolerance or fall below required confidence levels

People-Augmented™ AI Solutions

DynamicOCR continually improves upon itself using our proprietary algorithms – making itself smarter every time a page is processed. In the event DynamicOCR detects errant data or the High-Value Business Data falls below a predefined confidence level – creating an exception -CAPSYS can route the case work in question back to your Knowledge Workers for their review. If the Knowledge Worker determines that DynamicOCR may have mis-identified relevant High-Value Business Data, the Knowledge Worker directly from the CAPSYS Web Client interface can affect the needed configuration change within the context of the existing case work in question. Upon the acceptance of the change by your Knowledge Worker, DynamicOCR will forever remember the correct position of the data to extract on a going forward basis. No CAPSYS System Administration functions are required to affect a change to DynamicOCR’s learning set. However, a Knowledge Worker must be granted the appropriate privilege by your CAPSYS Administrator to make a correction to any errant data. 


Allowing your key Knowledge Workers to participate in the learning cycle of DynamicOCR contributes to DynamicOCR becoming smarter over a short period of time, thus minimizing your Knowledge Worker’s interaction with errant case work. Empowering your Knowledge Workers with the ability to accelerate and enhance DynamicOCR’s intelligence, reduces or completely eliminates having your CAPSYS administrator stop production so he/she can affect any manual configuration changes to DynamicOCR. 

Smarter Every Day

DynamicOCR finds your key, High-Value Business Data automatically and does so in just seconds. DynamicOCR operates the Cloud and can easily scale based on your workloads. DynamicOCR gets smarter every day as your documents are processed through our engine because of its built-in, self-learning technology. The more you process through, the more DynamicOCR contributes to its learning set and the smarter it gets. DynamicOCR’s results can also be validated against a database containing prior authorization numbers, loan numbers, member data, etc. using CAPSYS Technologies’ DbPRO™ ExecuBOT - further improving and enhancing upon DynamicOCR’s accuracy while contributing to its learning set. 


In summary, we believe theDynamicOCR can bring a substantial productivity enhancement to your organization. Contact CAPSYS Technologies today for more information or to schedule a demonstration. Watch a brief demonstration by clicking here!

Share by: