We accomplished these projects for various consulting, integration, and vendor organizations. Many relate to insurance applications in life, health, and property and casualty environments. The solutions, however, pertain to many other document and workflow intensive industries, including financial services and government.
Implementations have frequently been in VisualAge/Smalltalk, as this is the primary development language of many of these companies. Smalltalk provides superb domain (business, document management, and workflow) modeling capabilities, performs nicely, and interfaces easily with client/server, legacy, and graphical interface environments. The application designs and object models can, in most instances, also be implemented in the various object-oriented and object-based vendor-specific languages being used at other companies.
Case Assignment¾ Automated Rule-Based Work Distribution
Auto-Foldering¾ Automated Rule-Based Document Foldering
Multi-Valued Attributes¾ Enhanced Enterprise Index and Search Facilities
Enterprise Relational Indexing and Search Engine
Work Introduction¾ Automation of Manual Exception Processing
Work Folders¾ Combining Document and Task Management
Auto Search Engine¾ Automated Rule-Based Document Retrieval
Personal Document Repository¾ A Workstation and File Server Facility
Document Import and Indexing Facility¾ Integrating and Unifying Tools
Life Cycle Management¾ Controlling New Work-in-Process Documents
Virtual Foldering¾ Rule-Based Folder Dynamic Folder Creation
Application Desktop Object Interface¾ Supporting Programming Teams
Case Assignment¾ Automated Rule-Based Work Distribution
Case Assignment is a wonderful workflow application that was designed and implemented for a life insurance underwriting department. This application saved hours, reduced cycle time, and improved underwriting effectiveness by assigning incoming cases to the right person.
This application assigns new underwriting cases (electronic folders) to an underwriter based on:
This application operates in either interactive mode or "auto-assign" mode. It consolidates pending work assigned across multiple workflow domains and updates the case folders with the newly assigned underwriter. The workitem folder is then available in the assigned person’s queue.
This application was implemented using Eastman Software Open/Workflow in a multi-domain NT environment. The programming was accomplished in VisualAge/Smalltalk and C. It represents an excellent example of an object-oriented application, since the Workitem objects being assigned encapsulate data from the workflow system, other legacy data sources, and collections of related policy folders in the queue. Similarly the Staff Member objects encapsulate properties and methods from many sources for schedules, authorities, skills, geographic coverage, as well as accessing the consolidated work pending for each individual across multiple workflow domains. The interactive interface uses object signaling (events) so that changes made to staff member attributes immediately result in visually updated assignment availability for the current case.
Auto-Foldering¾ Automated Rule-Based Document Foldering
Auto-foldering is a heavy-duty robot application that places newly arriving (usually scanned) documents into their proper folder, and ensures the folder is subsequently placed in the correct workflow queue.
The robot runs unattended as a server process and uses rules to specify what folder should be the target for a particular arriving document type. If the target folder is currently locked (in-use), the arriving document is queued for subsequent foldering retries. In the folder is currently not in workflow, it uses rules to determine if the folder should be placed back in play, and if so, to which queue it should be sent.
This application has been implemented for Eastman Software Open/Workflow in both NT and OS/2 server environments for property and casualty, as well as life insurance organizations. It provides a powerful range of customization options, and since new documents are foldered immediately upon their arrival, it can out-perform standard rendezvous queues.
Multi-Valued Attributes¾ Enhanced Enterprise Index and Search Facilities
Many document management indices can only store a single value for a document or folder attribute, such as policy number. A great many documents for both underwriting and policy service applications pertain to multiple policy numbers. We designed and implemented an enterprise-wide multi-value indexing facility to augment the vendor’s indexing and search facility.
It was another good example of an object-oriented application, since the split behavior of indexing and searching could be performed in Item methods that were inherited by both Documents and Folders. The application programmer simple provided a collection of policy numbers, and the object updated both the vendor’s system (using APIs), and the SQL tables. Similarly, searches were federated across the two indexing facilities. This application’s design was complicated by the fact that the multi-valued attributes themselves may be composites of different elements, and multi-valued attribute ordering is often important.
This application was implemented using MVS DB2 tables, C, and VisualAge/Smalltalk for use on clients and servers. The document management system was IBM VisualInfo, although the same application is pertinent to other vendor packages. The multi-valued attribute application could also be translated to languages that don’t support inheritance by using interfaces.
Enterprise Relational Indexing and Search Engine
Subsequent to implementing the Multi-Valued Attributes, We designed and implemented a enterprise-wide document and folder indexing facility for a property and casualty insurance company which provides a single attribute index for multiple document management systems and/or multiple independent server domains.
This facility allows a single SQL query to be executed to retrieve documents that are stored in multiple heterogeneous or homogeneous document management systems. For example: a mixture of IBM VisualInfo, Eastman Software OPEN/Image, or Lotus Notes systems.
This design must be robust and comprehensive, since it requires all documents and folders to be consistently routed to a process that will update the central index. Luckily, vendors now have excellent workflow queues and vendor independent enterprise message queuing tools like IBM’s MQ/Series.
The concept of such independent indexing is quite consistent with the direction the Internet is taking us. Here independent indices, using search engines like Yahoo and Alta Vista, provide access to documents that are stored in systems quite foreign to the indexing system. Indeed, we have adopted other useful Internet Engineering Task Force (IETF) concepts, such as Universal Resource Identifiers and Names (URIs and URNs), to allow the application to retrieve the documents without knowledge of their specific location.
Work Introduction¾ Automation of Manual Exception Processing
Many manual workflow processes are triggered by exception and scheduling reports produced by legacy applications. We designed and implemented a work introduction subsystem to capture these paper-targeted tasks and create electronic workitems that are routed by a workflow system to appropriate personnel.
The application was divided into components to make maintenance easier. Workitem transaction definition programs obtained the basic information for each transaction, often by parsing the report print files. These programs were subject to frequent changes and additions as reports were modified. A workitem generator program read the output of the workitem transaction definition programs and created actual workflow items which were routed for processing according to the workflow route definition.
This application was implemented for Eastman Software Open/Workflow on an NT platform.
Work Folders¾ Combining Document and Task Management
Many insurance service and underwriting transactions require that functions be performed at both at both a macro- and a micro-level. For example, Harry marries Sally and wants to change his beneficiary and borrow on his policies to buy a house. He sends a letter to his insurer with this dual request.
In a manual workflow world, this might result in either the Service Team (which handles bene changes), or the Disbursement Team (which handles loans) getting the request, and then photocopying it for the other team. Harry gets two letters back in the mail and the runaround should he try to find out the status of both requests on a single phone call. Each team, in turn, may have to update multiple legacy systems, because the policies were taken out years apart, and different administration systems handle the historical differences introduced during each decade of insurance business.
Work Folders simplify this problem. The folders are associated with a Task or Activity List that defines multiple workflows and multiple transactions (for each policy) with each of these workflows. The single folder allows the status of the case and all documents associated with it to be viewed together. The multiple macro workflows (in this case Service and Disbursements) can either flow in parallel or serially, as the workflow software and its rules permit. The multiple micro transactions (in this case different policies to be changed in their respective admin systems) are often accomplished by the same person at a workflow step. This specialist can now update the status of each activity (since some admin systems may be online and others overnight batch) and get credit for everything he or she accomplished. When Harry calls, the customer service specialist can relay exactly what’s been done and see any associated requirements or response letters.
We have designed Work Folder applications for both life service centers, and property and casualty renewals and endorsements for Eastman Software Open/Workflow and IBM VisualInfo, respectively. The latter was implemented using VisualAge/Smalltalk.
Auto Search Engine¾ Automated Rule-Based Document Retrieval
As part of the life cycle of Work Folders, it is often necessary to fetch documents that are related to the transaction being processed. For one of my clients, we created an Auto-Search Facility that gathered relevant documents to populate the work folder according to predetermined business rules.
For example, to process an automobile renewal for a commercial customer (we’re talking small fleets now), the business rules would be to gather any claim documents for the past 12 months, motor vehicle and driver reports for the last 12 months, client financial documents for the last 24 months, any customer correspondence for 14 months, and any other notes or documents that were marked for the automobile line of business for 24 months.
A graphical rule definition facility allowed these rules to be specified for any automobile renewal. Other rules were defined for other lines of business. After the folder was created, it was routed to the auto-search workstep where a server process selected the appropriate rules and populated the folder with the resulting search documents. The documents themselves were then to be queued for staging to magnetic storage.
This application was implemented in VisualAge/Smalltalk for IBM VisualInfo. The same approach would work, of course, in other languages and with other repository systems.
Personal Document Repository¾ A Workstation and File Server Facility
In a wide-area network—and particularly an Internet—environment, it is becoming increasing useful to be able to define a repository for work-in-process documents and folders on a user’s workstation. This affords the ability to do useful work when the mainframe, a server, or a communications link is not in service. It also allows users to conveniently manage folders of documents when working on the road or visiting clients. For developers, is also handy for proof-of-concept as well as marketing demos.
We designed and constructed a workstation and server document management repository that stores documents and folders, as well as their attributes on a standard file system directory. The programming objects for this repository are polymorphic to the objects for the vendor’s repository, which means that they respond to the same messages to create and delete items, add and remove contents of folders, launch viewers, and index and search for items.
My client elected to store the user’s personal document directories on a file server, rather than on the workstation, so that these files would participate in daily backups.
By assigning Universal Resource Names to these items, it is possible to define folders that contain a mixture of locally stored documents and host stored documents. A migration scheme allows documents to be moved from the workstation to the host repository for sharing or permanent archiving. Thus new documents can be created, imported, or scanned at the workstation, fully indexed at the workstation, and then migrated to the host document management system at a convenient time, such as when the traveler connects back to the network, or when work on a case or project is complete and the documents need to be shared with others. We envision this facility as the bottom tier of a three tier document management hierarchy where a groupware product serves as the second tier for sharing documents in the department, and the third tier is an enterprise document management system for permanent archiving on optical and sharing documents beyond a department.
This facility was also used by my client to provide desktop object persistence on the user’s underwriting workstation. Tools, documents, folders, preferences, and other data associated with the graphical desktop is persisted as the contents of a desktop folder on either their workstation or a shared file server.
This facility was implemented in VisualAge/Smalltalk. The object model could be migrated to other object languages.
Document Import and Indexing Facility¾ Integrating and Unifying Tools
A rather spiffy desktop facility was designed and implemented to import and index new documents from several common sources in a unified manner. The document source could be any one of:
The targets were the enterprise customer document management folders, which were represented as icons on the user’s desktop. By unifying these import facilities, the user could multi-select documents and directly add them to a customer folder. Individual attributes, such as document description, could be rapidly changed. For the user, this is all very fast and convenient.
All of the import sources were implemented in VisualAge/Smalltalk except the Internet Browser, which was designed but has not yet been programmed.
Life Cycle Management¾ Controlling New Work-in-Process Documents
In production environments, documents need more than the simple protection provided by a check-out/check-in facility. We designed and implemented an extension to a vendor’s document management system to provide the following additional document states:
This facility was implemented in VisualAge/Smalltalk using the IBM VisualInfo repository.
Virtual Foldering¾ Rule-Based Folder Dynamic Folder Creation
Many organizations organize their documents by well-defined categorization. Certain document types belong in clearly defined customer or policy folders. This well-defined filing structure, by the way, is actually the enabling scheme that allows the Auto Foldering robot to operate.
When using a custom folder retrieval and viewer application on the desktop, it becomes possible to eliminate the entire effort of creating these folders and incrementally storing documents in them. Using appropriate search and filtering logic, the application can dynamically assembly a collection of documents of particular document types from the attribute indexing table.
We designed and implemented this concept for a property and casualty insurer using an object-oriented approach. Model folder objects are created based on search specifications. The contents of these folders are identical to what would have been obtained by traditional physical foldering. These model folder objects can be used by both desktop visual objects, as well as server applications. The existence of these virtual folders, of course, does not preclude the use of standard folders to satisfy work folder requirements and ad hoc foldering requirements.
This was implemented in VisualAge/Smalltalk using IBM VisualInfo. The same approach and object model can be used with other languages and vendor products.
Application Desktop Object Interface¾ Supporting Programming Teams
Programming teams that are developing robust document- and workflow-enabled applications, such as underwriting or service desktops, require an easy-to-use interface to the supporting vendor products and their APIs.
We designed and developed object interfaces to various vendor products to encapsulate the details on a vendor’s implementation into a convenient domain model. This allows non-document management trained programmers to utilize these facilities to create, persist, and retrieve information with vendor-independent object interfaces.
Among the features of these object sets are:
These models have been used with Eastman Software OPEN/Image and OPEN/Workflow, IBM’s VisualInfo software, as well as our vendor-independent workstation document persistence model. The objects were implemented in VisualAge/Smalltalk and are portable to other languages.
[Home] [What's New] [Links] [Services][Site Table of Contents]
Copyright © 1997-2000 by Gary M. Gershon. All rights reserved.
Last Changed: 08/19/00
Comments to: Gary Gershon (Gershon@Celsus.Net)