SAN MATEO (03/13/2000) - The ability to use computer-aggregated data to discover patterns and critically analyze business processes is one of the important benefits that computers have bestowed on business. Doing business on the Internet allows companies to gather real-time data, process that information immediately using business intelligence tools, and react accordingly.
Neugents is business intelligence technology from Computer Associates International Inc. -- built to work with Jasmine ii -- that can enrich custom applications by extracting business meanings from a stream of data. During our recent evaluation of Jasmine ii, we took an early look at Neugents as well, and found it easy to integrate with existing or new applications. However, it requires some data manipulation, and has an unappealing, limited GUI.
To understand the benefit that Neugents could provide, we have to consider how the speed of e-business operations outpaces our ability to respond appropriately when problems arise.The recent spate of DoS (denial of service) attacks is a good example of this type of business challenge. For the companies targeted, each moment that their service was affected was potentially damaging to customer relations and revenue. It is reasonable to assume that nonmalicious events, such as a sudden convergence of visitors or a system malfunction, could produce similar havoc. Problems happen fast, and companies need to be able to respond with similar efficiency. Injecting some business intelligence into your systems can help you avoid responding too slowly.
Neugents was first released in December 1998 as part of Unicenter TNG, CA's flagship management software. In that context, Neugents is used to make predictive analyses of systems performances or equipment failures. Jasmine ii's Neugents discovers cause-and-effect patterns among data elements and makes predictions based on those patterns. These predictions can then be used to alert operations managers to unusual situations or simply to make regular business projections such as inventory needs.
Neugents can be used in a variety of business contexts, but it requires some previous data analysis and manipulation. Its agents analyze two distinct sets of data: an input flow and an output flow. Because the product finds relationships of cause and effect between the two flows, you have to organize your data in input and output flows. For example, in a Web server performance analysis, the number of online users would be the cause and the response times the effect.
If text is part of your data, you need to do some more work, because Neugents only understands numbers. This preparation activity can be significant if you are dealing with large amounts of data.
Once you've formatted your data properly, Neugents scans it and creates a summary view showing how changing values in input fields affects the results.
At this stage, you can create a profile of your data, which Neugents can use to validate future predictive analysis.
If you'd like Neugents to help you predict how changes in input variables affect output results in your data model, you'll have to feed Neugents a new input stream. Neugents will then validate the new data against the profile and make predictions according to the rules it had already learned about your data.
Neugents' agents make a quality assessment of the new flow of data and express a numeric rating to indicate how the new data matches the profile. The agents also associate a probability percentage for each output value, which will help business analysts put a value on the believability of the forecast.
Probably the most disappointing aspect is a poorly executed wizard-based user interface, which should've made it easy for us to feed the product data, but didn't. On the positive side, developers should find it very easy to merge this technology with their code thanks to a monolilthic set of Dynamic Link Library modules.
This implementation of Neugents allows developers to apply intelligence to the many data sources that the Jasmine platform supports, and should have a tangible impact on your company's ability to respond at the pace of e-business.
Mario Apicella (firstname.lastname@example.org) is a senior technology analyst for the InfoWorld Test Center.
THE BOTTOM LINE: BETA
Business Case: Using a pattern-recognition technology such as Neugents in an e-business environment can increase a company's understanding of the factors affecting business performance and ultimately improve decision-making.
Technology Case: Neugents' self-contained classes are easily deployed within Jasmine or in custom applications.
+ Generic -- can analyze any cause-and-effect relationship of data+ Creates trainable and verifiable models+ Flexible, modular architectureCons:
- Poorly executed wizard-based GUI
- Requires considerable data formatting
Cost: Not available at press time
Platform(s): Any Windows programming tool that can make calls to dynamic link librariesShipping: Summer 2000Computer Associates International Inc. Islandia, N.Y.; (800) 225-5224; www.cai.com.