Transportation company Allied Pickfrords is using enterprise capture in Australia as a way to deal with an ever-growing volume of business-critical information generated both internally and externally, using a Datacap solution deployed by IBM FileNet reseller Elementum

Garry Stephenson, Director and Principal Consultant at Elementum, said the company has implemented a range of enterprise capture solutions for Australian organisations including Iron Mountain, Synchronised Software (SyncSoft),Nissan, Ullrich Aluminium, Henley Properties and Allied Pickfords.

Elementum’s heritage is across the broader information management realm including records and knowledge management however it specialises in document imaging and automated data capture.

Allied Pickfords was Elementum’s first Datacap client in Australia. Parent Company SIRVA was a reference site for Datacap in the US.

Allied Pickfords has over 40+ sites around Australia where documents are scanned using a Multi-Function Device (MFD).The resulting images are sent via email for centralised capture at the Allied Pickfords Head Office in Dandenong, Victoria, using a custom Taskmaster Application.

The application receives all incoming email and converts attached documents from various formats (PDF, Word, JPEG) into a tiff file format.

Taskmaster auto-classifies by a combination of techniques (Barcode Id. Fingerprint, Key Word/Phrase Search (40% OCR)) then captures and validates key data e.g. Removal ID, Client Surname and location.

After performing a database look-up to validate customers and ascertain a destination for Images, "named" PDF files are then created for each document required. These are then exported to a destination folder within Allied Pickfords’ core business system.

Banking Solution

Elementum has also implemented a new Datacap Capture system for Iron Mountain, on behalf of one of the four major banks that is introducing Automatic Document Classification for Certificates of Title, Mortgages and many other document types associated with buying and selling a property.

“Of 107 Business Lending Documents to be classified, 20 individual document types made up 95.8% of the volume and of 234 Retail Lending Documents to be classified, 20 individual document types made up 87.8% of the volume and 40 made up 95.1%” said Stephenson.

For this nationwide solution, loan documents are imaged in the state of origin and then images are transferred for centralised classification. The classification solution provided by Elementum using IBM Datacap Taskmaster again applies a number of different identification techniques: Barcode Recognition, Fingerprint and Key word/phrase search.

“One of the major challenges in using Key Word/Phrase searching across unstructured documents is uniqueness of the target.

“By analysing your document sets and identifying where the target may be in a specific area then restricting the OCR Capture range to say 30% you are less likely to incorrectly classify documents with the same target phrase used in different context later in a document. Lessening the OCR area also provides productivity improvement.” says Stephenson.

The bank has remote access to a Taskmaster based QA function through Taskmaster web via a secure Single Sign On process. This functionality provides for the bank to manually classify unidentifiable documents (sometimes) erroneously forwarded for capture.

For any Auto-classification application, Elementum’s Stephenson recommends beginning with a limited set of documents to avoid biting off more than can be managed first up while also delivering faster ROI.

“Experience has shown that whatever the type of documents that are being classified, generally 20% of the document types being scanned can end up making up 95% of the volume. So, rather than setting up thousands of templates or rules to account for every different document type, it’s better to concentrate on those that will make up the bulk of the volume. In the end, it can easier to use a human in many instances of low volume document types, rather than setting up countless classification rules.

A recent review of documents undertaken for an organisation within the Health Service industry revealed; “of 1500+ form types currently being processed, 17 made up 27% of the total number imaged,” said Stephenson.

“When rolling out this type of Document Classification project it is important to get an understanding of what the makeup is and focus first on the 17 that make up the 27%, and then further analyse and roll-out in a phased approach.”

The tyranny of distance is well understood in a country with the geographical challenges of Australia. According to Stephenson, distributed (image) Capture supported by centralised Recognition and Data-capture with either centralised or decentralised verification/correction provides the platform required to meet those challenges.

“The immediate benefits of Distributed Capture are obvious and tangible: Timeliness of receipt (immediate versus 1 or 2 days); and cost savings associated shipping of hard copy.

“Beyond these obvious benefits there is a further, far greater benefit to be realised in terms of both cost and operational efficiency. The centralisation of verification/correction provides for:
- a lesser number of full time trained resources working more hours on specific applications
rather than many part timers across state based operations; and
- Greater redundancy in the workforce with a centralised knowledge base.

“Apart from the benefits listed above, some applications (lending) actually mandate that hardcopy documents (e.g. Certificates of Title) remain in the State of Origin for subsequent access.”

“It is essential to thoroughly understand your application and desired outcomes and complete a detailed requirements specification before undertaking a proof of concept (pilot) to provide insight into solution. A key point to remember when designing a capture solution is that it is generally 10% technology and 90% process.”

Before you begin …
■ Test suitability of technology against a cross-section of document
classes
■ Validate and Refine Operational Processes
■ Establish Business Rules
■ Benchmark Productivity across all key functions
■ Validate output file sizes for capacity planning
■ Where possible; when piloting, go end-to-end including uploads