An old adage states that 25% of MIS department work relates to catalogs. Today e-marketplaces - or Net markets - are flooding virtually every business sector as more and more companies scramble to migrate their supply-and-demand-side operations to the Internet. Product and catalog integration is a vital part of these moves.

The idea behind online marketplaces is to create an electronic environment where nimble buyers and sellers can effortlessly move in and out, exchange information, and ultimately grow their businesses. The obstacle, of course, is standardization and integration.

Participants are faced with Net market vendors who rely on proprietary systems to ensure loyalty and adoption. If a business wants to participate in several Net markets, the challenge is to understand how to use information from each. When companies have to publish business information and data, dynamically update the content, and then integrate it with various business flow applications - using more than one standard - joining the e-business train may seem more like riding a roller-coaster.

XML Makes Matters Less Daunting
In addition, catalog information isn't just about marketing products, but also about the need to marry into specific back-office requirements, such as accounting and inventory management. On top of this, enterprises are now asking Net markets to provide features directly affecting catalog management, such as interexchange operability, robust collaboration, and demand forecasting. Fortunately, with the rapidly increasing acceptance of eXtensible Markup Language (XML) as the standard for e-business, these demands aren't as daunting as they once seemed.

For developers working with XML and catalog management logic, the ultimate goal was to create a solution that would facilitate multiple catalog management paradigms, real-time information exchange, end-to-end transaction integration over the Internet, and complete interoperability with all back-office systems. Of course this isn't a new idea, and EDI and a variety of ERP system vendor solutions have been targeting this for years. Recently the Open Applications Group (www.openapplications.org) produced XML exchange formats and version 7.0 is now available. As a broad open standard, however, the OAG work is limited by its ERP and COBOLesque heritage and backward compatibility needs.

The large Net market vendors have not been slow to address these areas. Corporations such as Ariba with cXML and Commerce One with xCBL have developed their own proprietary XML-based vocabularies. These specific solutions may be workable for certain industry verticals being served by a particular company. For cross-industry communication, however, or interexchange operability, these unique vocabularies are limited.

Some companies such as POET software are exploiting this need (in the short term) by producing XML integration server platforms. Indeed, one of the most important issues facing XML today is the need for a rule set outlining how XML can be used ubiquitously for e-business. The closest we have to that is the major work being done by ebXML (www.ebXML.org) to develop technical specifications for using XML to effect interoperable semantic structures.

This article looks at the initial work on utilizing these concepts to produce an open and neutral XML catalog exchange format, the CatXML implementation.

The DLA EMALL
The CatXML Approach
CatXML is a public open software solution developed by a team of XML consultants to the U.S. Government Defense Logistics Agency (DLA) EMALL project, aided in part by XML Global Technologies, Inc. (www.xmlglobal.com), who have implemented a live server site.

The DLA EMALL requirements are targeted at small businesses with minimal e-business resources that provide a significant level of service to the U.S. armed services. Specifically, the EMALL project is designed to offer the four services (Army, Navy, Air force, and Marines) a single requisition system for common parts and replenishable supplies.

Two proof-of-concept systems were created to demonstrate how small vendors can be quickly integrated into a larger central system, and to provide a distributed information grid that can supply real-time catalog information without prohibitive performance overhead. From the original base syntax, XML Global has now developed an advanced version that supports completely scalable dynamic information exchange for CatXML compliant points-of-presence. A flexible "smorgasbord" approach allows industry partners to implement matching profiles quickly and effectively with both inbound and outbound information exchange supported and the information server approach as the foundation.

The advanced version of CatXML is positioned to take advantage of emerging XML technologies, including dynamic schema methods (www.w3c.org/RPT/schema), open interchange specifications from the ebXML Initiative (www.ebXML.org), and leading catalog interchange formats from industry associations such as the Open Applications Group (www.openapplications.org). The goal is to provide a vendor-neutral specification founded in XML syntax that can be freely implemented.

Since XML Global and the DLA are not vendors of proprietary catalog services or solutions they have no particular ax to grind. This has enabled the creation of a "best-of-breed" approach to developing CatXML syntax and semantics. Catalog developers thereby gain the advantage of using proven component and method designs, while locally customizing them to their specific industry base. The potential payoffs are the interoperability and cost savings across a broad spectrum of marketplaces that are delivered by the flexibility and consistency of CatXML (see Figure 1).

Let's go to the XML
Before looking at specifics of the XML we need to look at some basics of organizing and structuring metadata. Too often we see metadata provided as an amorphous blob of elements and attributes with little regard for using parent/child structures to create groups of related elements and thus, easily accessible information blocks. Once a well-defined group structure is established, a higher level of organization can be created by separating the XML itself into physical components included in a controlling parent template. This allows trading partners to be flexible about the components they select and the versions and subdefinitions they apply to the master templates. All these techniques can be found in the CatXML syntax examples (see Figure 2).

Next, CatXML seeks to address the problem of naming and semantics. Ultimately the ebXML Registry systems based on reference Unique IDs (UIDs) will be the implementation of choice as they become available in 2001. However, the earlier work on CatXML uses the term Bizcodes and QIC references for these linkage references (see www.bizcodes.org). Linkage references allow collaborating businesses to synchronize their business semantics by labeling each piece of interchange information. Therefore the context and meaning can be readily equated to similar usage within the receiving business systems. Similarly the CatXML interchange transport support is currently centered on an Access/Content_Input/Response_Output model (see Figure 3).

Again, ebXML is actively developing transport formats for Registry interchanges along similar lines and eventually, once these are approved release candidates, the current CatXML model can be migrated to support them. The value of good semantic models cannot be overemphasized here. Well-implemented XML should provide the foundation and stepping-stone to emerging standards.

To round out this section here's a sample of the actual XML syntax. more details on this and examples of DTD and XML document instances can be downloaded from the CatXML web site (www.catxml.org) along with documentation and examples (see Listing 1).

The intention of this section was to briefly introduce ideas and concepts that can be studied and developed further. The next section rounds out this review by considering the broader deployment needs.

Deployment Considerations
Thus far XML Global, the XML Fund, and a selection of commerce partners have deployed a first implementation of CatXML utilizing the GoXML information repository system. The strength of the CatXML design is its ability to support any legacy information structure while simultaneously providing completely dynamic output format. In addition, because the CatXML architecture supports a distributed query information grid model, it avoids performance bottlenecks, allowing small trading partners to expose an open interface to a broad selection of online Internet marketplaces.

The simplest CatXML-compliant implementation model is a low-cost point of presence (CatXML POP) solution. The most obvious feature of the approach is that any computer system, anywhere, can access the catalog anytime. As a business model this is well suited to many small suppliers who simply want to enlarge their market presence. More sophisticated operations, however, are required to restrict and manage trading partner relations, and so wish to define access channels and profiles for each of them.

The CatXML interchange server (CatXML IS) provides the next level of functionality by allowing two or more trading partners to share information in a closed model using locally custom-defined information formats. A typical model would be a supplier with three partners, each using different line-of-business (LOB) systems that export information into fixed formats individually mapped to a CatXML-compliant format. To deliver this kind of enhanced functionality, the CatXML specification provides full interchange semantics. This enhanced performance includes a scheduler to automatically acquire information from remote suppliers. Typically this would be a daily or weekly event, timed to coincide with catalog updates from the vendors' business systems. These updates are then consolidated into the existing catalog information. CatXML also supports the use of channels within information folders, so only content that matches a partner's retrieval profile will be returned to a remote system.

Following this is the CatXML service provider (CatXML SP) where individual partners contract to a central service that allows them to manage and control their catalog information distribution. Performance is a major concern in this model, as is support for real-time information access on items such as price and quantity on hand. The CatXML syntax provides a flexible model that supports real-time information querying on an as-needed basis. The CatXML information grid then follows as an extension, linking a combination of the other CatXML models. Large organizations such as the U.S. Government EMALL system require these kinds of functional business capabilities.

The CatXML Initiative is unique in several ways. It uses XML content in a truly native fashion without requiring participants to alter the format of their own local XML content. In addition, it exploits the capabilities of XML syntax to provide a high degree of maintainability and extensibility in line with the original design goals of XML itself.

Previously organizations had been capable of sharing and reusing business objects within their own confines, but very rarely achieved re-use across business partner systems. Now the new business models being created by widespread content sharing on the Internet are requiring that organizations share content at unprecedented levels. The ebXML Initiative is also focused heavily on the need to support such capabilities.

These technology advances translate into specific business benefits. The ability to reach a large base of small vendor partners is one model, as evidenced by the U.S. Government's EMALL Initiative. Another model is the ability to reach a widely varying supply chain base where a large company is sending catalogs to thousands of locations, all of which require specific customization.

A further model is to deliver to a grid of distributed content points. Each point can respond dynamically to requests, so performance can be optimized. Static information can be cached at central locations and periodically updated, while transient information such as stock levels can be directly queried in real time as individual content points.

CatXML is delivering on the promise of XML to provide open interoperable information content.

References

1. Internet-based selling web sites: www.freestore.com, www.vista.com, www.commerceone.com, www.clubselling.com
2. Legacy Electronic Data Interchange (EDI): www.disa.org
3. Bizcodes: www.bizcodes.org
4. EDI - Electronic Data Interchange: www.disa.org
5. The CatXML reference information: www.CatXML.org
6. The ebXML Initiative: www.ebXML.org
7. GoXML: http://goxml.com

Acknowledgment
The auther wishes to thank Bruce Peat at eProcessSolutions for his help with this article.