February 1996 € Volume 23 € No. 1

Cutting Edge

By David Blanchard

National Library Develops Intelligent Index

The U.S. National Library of Medicine (Washington, D.C.), the world's largest medical library, has developed an intelligent medical indexing system called MedIndEx (Medical Indexing Expert). The library has over 4.5 million holdings (including books, journals, reports, manuscripts and audio/visual items) that are accessed online by health professionals worldwide. This information is used to help diagnose illness and disease, as well as for research on particular biomedical topics.

The ongoing indexing of the library's resources is a vital but time-consuming task. According to Susanne Humphrey, an information scientist with the Lister Hill Center, a research division of the National Library of Medicine, the library decided to develop an application that would transform the existing data entry system into an intelligent knowledge base that combines artificial intelligence and information retrieval principles. Her idea led to the development of MedIndEx, an intelligent knowledge base.

With the conventional system, indexers typically had a number of windows on their screen at any one time. This allowed them to simultaneously view several different applications or systems that provide data entry facilities, and a retrieval system that can access the library's medical thesaurus and other relevant databases. The MedIndEx prototype introduces a new type of data structure which encodes the combined factual and procedural knowledge established in the conventional system.
In addition, this knowledge base provides interactive, situation-specific assistance prompted by the users' interactions with the system. The software actually aids the indexer by suggesting or prescribing possible paths to follow in the process of indexing an article or set of articles.

EDS Turns to Virtual Costing

EDS Technology Architecture (Plano, Tex.), an international information technology company's research group, has developed a visually-enhanced cost technique called Virtual Activity-Based Costing (VABC), which helps the company understand how money is being spent at all levels of its operations. According to Steve Chenoweth, senior systems engineer at EDS Technology Architecture, with VABC and virtual reality software, management is able to take advantage of accounting processes to make effective management and decision-making processes.

Traditionally, accounting activities have focused on valuing an enterprise for financial reporting purposes. Costs are identified according to the category of the expense such as salaries, supplies and fixed costs. In VABC, costs are charged to the activities where the resources are actually consumed, allowing the identification of the causes of costs. While this generates much more information than traditional accounting, this information is critical to understanding and improving business processes.

This additional information needs to be captured, organized and presented in a fashion that facilitates comprehension and decision-making. In the VABC application, each activity is represented by a vertical bar whose height indicates the cost and whose color depicts the value (green is high, yellow is medium, and red is low). Surveying this 3-D landscape, the user can determine what activities warrant further exploration.

One of the first proof-of-concept applications was for the data visualization of credit union data. The system first shows data displayed in a traditional manner, and then multi-dimensionally using 3-D visualization. Scanning through the 3-D horizon, the user sees information in a new way.

Presenting and navigating through this complexity with an interactive, 3-D user interface allows users to zoom in and out, navigate up and down the hierarchy, or drill down to the lowest level of detail while maintaining a perspective of where they have been and where they are going.

Wrangler Replenishes with Neural Networks

Wrangler (Greensboro, N.C.), a manufacturer of men's and boy's jeans, shirts and knitwear, is using neural network technology in a reengineering effort to increase sales volumes, lower inventory investments, and improve inventory turns for retail customers and for the company itself.

An important part of supply chain reengineering at Wrangler has been revamping product forecasting and production planning. By forecasting production needs more accurately, Wrangler will be able to maintain a high level of in-stock customer service while carrying less finished goods inventories.

Most consumer goods manufacturers plan production based on forecasts of order demand, and Wrangler is no exception. However, Wrangler's order demand was inconsistent with sales. To improve production planning and forecasting, Wrangler has begun using neural network forecasting technology. It generates forecasts based on consumer demand data, rather than retail buyers' orders, to drive production planning.

This information is combined with consumer sales information to feed a neural network-based forecasting model. Forecasts are created for each retail chain. The aggregate of those chain forecasts now drives Wrangler's production planning. Wrangler's manufacturing and sourcing are now matched up with actual consumer demand.

Contributing editor David Blanchard is editor of Intelligent Systems Report, (216) 677-4210, E-mail: blanchard@lionhrtpub.com.

E-mail to the Editorial Department of OR/MS Today: orms@lionhrtpub.com

Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com