Burlington Northern Santa Fe Corporation (BNSF), headquartered in Fort Worth, Texas, operates one of the largest railroad networks in North America, with 33,000 route miles covering 28 states and two Canadian provinces. BNSF has a long list of clients who use BNSF's railroad network as a primary means of transporting goods in North America. Most of these clients today interact with BNSF using the rail industry's EDI404 Standard for communicating with BNSF for sending in Bills of Lading (BOL), etc.

However, BNSF still has clients who use the traditional facsimile method for sending in BOLs to BNSF. On an average BNSF today receives 1,400 to 1,500 faxes of BOL per day from various customers. The BOLs faxed by BNSF's various clients do not follow any specific standard format and each BOL has a different format for not only each client but in some cases the same client could have a different format for its BOL depending upon the location of the particular client.

Upon receipt of the faxed BOL, BNSF uses manual data entry services to key in the data to an EDI system which produces an EDI404 transaction that is then used by their waybill generation system running on their mainframe system to complete their revenue billing. During the manual process if the data in the faxed BOL is unreadable or missing, the data entry staff will contact the responsible client to get the data, ensuring that a certain service level is maintained. If the data entry staff is unable to get the information from the client or is not able to validate the information quickly enough, these transaction details are then passed onto an exception handling group to do further research. The data entry staff proceeds with the rest of the data entry and validation process of other lading bills leaving the partially finished EDI transaction set as it is.

As a process, based on the BOL received, BNSF has a window of two hours or less for regular BOLs and an even shorter window of 45 minutes or less for intermodal BOLs to generate corresponding waybills only after which the cargo can be loaded onto the railcar and dispatched. Thus, this is a very time bound process and any significant delay will amount to the cargo getting piled up on trucks lined up at BNSF's cargo loading stations.

BNSF wishes to eliminate manual data entry in the above process. The best solution would be to force all their existing clients to stop faxing BOLs to BNSF and instead start using the standard EDI404 interface to send BOLs to BNSF. However, BNSF understands that its clients will not be willing to make the transition overnight and expects this to happen over the next few years. In the interim period, BNSF plans to use cutting edge OCR/ICR technology solutions from CMC for automating this process such that BNSF receives a valid EDI404 document as a deliverable that can be then used by their existing system to complete their revenue billing in a timely fashion.

Problems in the current setup

BNSF faces the following main issues with the existing setup:

	With the manual data entry process for EDI 404 generation, BNSF has to depend on the accuracy of the data entered by their clerical staff from the faxed BOLs.
	Higher operating costs on the maintenance and upkeep of a 24 x 7 operation.

Solution background
BNSF wished to use cutting edge technology to automate the data entry process of multi format BOL data using OCR / ICR technology solutions from CMC. OCR / ICR technology has become an extremely important area where hard copy of data has to be converted after image recognition and processing into required data formats. BNSF leveraged these technologies to meet the following objectives:

	Recognise and process fax images using ICR / OCR technology
	Convert the same into required EDI format

In order to prove the concept, BNSF used a prototype approach where faxed BOL formats from a few specific clients were studied and processed using CMC's OCR / ICR technology based on the Freedom suite of products and the corresponding EDI404 transactions were generated. BNSF found a high consistent accuracy rate on the generated EDI404 transactions during this pilot run and based on this has decided to proceed with automating all the major clients of BNSF (about 200-250) who continue to send in faxed BOLs to BNSF as on date.

CMC's solution is deployed keeping in view the existing requirements, existing infrastructure as well as future plans of BNSF.

OCR/ICR technology
The OCR / ICR solution is a workflow system which consists of a series of components linked together. The system works with the central module. It is the interface to the run-time part of the system and is responsible for linking all its various components. Through the central module, you can activate several modules, according to defined permissions, thus controlling the flow of data in the system.

The OCR solution reads, understands and processes incoming fax images of the BOL without the need to predefine the structure of various faxes. The system works with the definition of keywords defined for identifying each field on the fax image. The system captures, validates and delivers the data from the faxed BOL to the required format. The OCR solution processes multi-page BOLs, which is kept intact in a batch file. Some of the features are:

	The system handles any type of Bill of Lading, including BOLs with complicated tables and multiple pages.
	The system's high-speed content recognition replaces the expensive manual process of keying in data from the invoice.
	The system recognises BOLs and the fields according to keywords and key terms situated anywhere on the page, such as shipper, from, sender, etc.
	Allows quality check of recognised characters and entry of unprecedented character recognition and error detection.
	Uses virtual engine comprising multiple recognition engines that vote to achieve the most accurate results.
	Converts and exports data that has finished the completion stage.
	Exports image files.
	Exports data to user-definable ASCII, INI or XML format or to a remote database.
	Delivers offers faster and more efficient processing.
	Delivers minimum human intervention to reduce labour costs and increased data integrity.

The ASCII-based file is then converted to EDI404 format using a utility developed by CMC.

Solution offered
CMC's proposed OCR/ICR solution for automating BNSF's waybill generation process comprises a fully integrated production system setup at its offshore facilities in India and uses a suite of OCR / ICR products from reputed OEMs.

A successful pilot project was executed for nine clients having about 30 different formats of BOL. Electronic copies of the faxed BOLs from these clients were transmitted to CMC in India for processing using OCR / ICR technology and then converted to valid EDI 404 transactions and transmitted back to BNSF. The solution was successful in achieving an overall recognition rate of 84 per cent in handwritten faxed BOLs and more than 96 per cent in machine printed faxed BOLs.

The OCR / ICR operations are as follows:

	The OCR / ICR application continuously checks for any fax image available on the server and any electronic faxed images available are pulled by the OCR / ICR application.
	The faxed BOLs images received on BNSF's server are transferred to CMC's server in India over a secure, dedicated 512 Kbps leased link.
	The faxed images are put into the OCR / ICR application based on the Freedom suite of products and customised by CMC for BNSF.
	The required fields from the faxed BOLs are recognised using OCR / ICR technology.
	The solution also contains validations and cross validations between the fields in a form to minimise the errors in the data.
	The valid ASCII-based data is then exported as final data and converted into the EDI404 format by a conversion utility designed by CMC.
	The EDI404 document is then sent to BNSF's EDI server in Fort Worth, TX over the same 512 Kbps secure, dedicated leased link.

During this EDI404 transaction generation process, the bills of lading are also validated using patterns stored in the database for automatically detecting and correcting any illegible data from the faxed BOL.

Before sending the mapped EDI404 strings to BNSF's server, CMC runs a pattern comparison to ensure the correctness of the EDI404 transaction. Once BNSF receives the EDI404 transaction from CMC, BNSF does a check for accuracy of the transaction based on which the transaction is processed and a waybill is generated against the same.

Service level agreements

BNSF and CMC have entered into stringent service level agreements for ensuring:

	Turn-around time as required by BNSF based on the type of the faxed BOL whether standard or intermodal
	Accuracy of the EDI404 transaction set generated against each faxed BOL

Benefits of the solution
An immediate benefit of implementing this solution from CMC is the reduction in the operating expenses in the waybill generation process.

On account of the stringent service level agreements with CMC, BNSF can expect high levels of accuracy and turnaround time and thereby get rid of wasteful cycles in the entire process of revenue billing.

Some of the intangible benefits include deployment of state-of-the-art IT solutions and systems for automating routine critical tasks like waybill generation and thereby being in the forefront in reaping the benefits of technology.