Enterprises are turning to the Web to deliver support information directly to end users. These self-service support tools, which provide access to data in Web pages, document stores and newsgroup archives, reduce the number of calls to the helpdesk and provide end users with a support environment independent of service hours and hold queues. However, many fall short of a Knowledge Base as defined above and do not incorporate new knowledge for reuse.

Metadata tools to build Knowledge Bases on the Web are making their way to the enterprise. Products from Informatica Corp.'s Metadata Exchange Architecture and SAS Institute's Data Warehousing hold information on how data is distributed across the enterprise on Web servers, multiplatform file servers, e-mail systems and databases, and describe it in different formats. For example, information on our open mail relay problem may reside in help files on a file system, in a multimedia-training workshop on an image database or in manual pages served up from a database. Metadata can bring these resources to bear on the problem by providing information about content, data type and location. Like a traditional Knowledge Base, metadata also can include usage patterns to indicate the information's worth by its use, along with policies and procedures applicable to information access and reuse.

Metadata repositories may be included with the various database servers, or there may be a separate repository, or even repositories, for the metadata. Standing alone, a metadata repository is a database that integrates heterogeneous data to include all possible metadata mappings between various data resources in the enterprise. Data mining tools then consult the metadata repositories to guide search engines or carry out data mining directly on the metadata to extract patterns and relations among data. As opposed to a static Knowledge Base of documents, metadata repositories are transaction-oriented systems with OLAP (online analytical processing) and data warehousing capabilities to continually mine data repositories.

Metadata can be the key to integrating disparate data sources. Building a metadata repository from enterprise resources and then mining the metadata to extract patterns can provide a virtual Knowledge Base for end users that makes available documentation, online help files, FAQs, hardware and software manuals, and online training. (See "E-Learning Branches Out," http://www.informationweek.com/826/elearning.htm.) For example, our NWC999 case solution can be represented in metadata, where a complete list of internal and external resources available for open mail relays could be detailed.

<rdf:RDF
xmlns:rdf="http://w3.org/TR/1999/PR-rdf-syntax-19990105#"
xmlns:dc="http://www.sendmail.org/antispam.html#">
<rdf:Description about="http://www.sendmail.org/antispam.html">
    <dc:Title>Preventing Relaying Through Your SMTP Port</dc:Title>

    <dc:Description> Mail must originate or terminate locally (on this host), with exceptions made for hosts listed, one per line, in /etc/sendmail.cR. (In this incarnation, this rule set does not allow relaying within your domain, except for the hosts listed in /etc/sendmail.cR. Note carefully, you must list fully qualified host names for each host you wish to allow relaying. Simply listing a domain name is NOT enough. Ideally, it would only restrict relaying from outside your domain. This is particularly useful if you are using mail clients such as MH or Eudora that like to try to find an SMTP relay on the local network to do outgoing mail submission.)

</dc:Description>

<dc:Publisher>Sendmail.org</dc:Publisher>

<dc:Date>2000-01-07</dc:Date>

<dc:Subject>

<rdf:Bag>

    <rdf:li>Sendmail 8.8</rdf:li>

    <rdf:li>Mail Relay</rdf:li>

    <rdf:li>SMTP</rdf:li>

</rdf:Bag>

</dc:Subject>

<dc:Type>World Wide Web Home Page</dc:Type>

<dc:Format>text/html</dc:Format>

<dc:Identifier>urn:NWC999</dc:Identifier>

<dc:Relation rdf:parseType="Resource">

<rdf:resource="http://purl.org/metadata/dublin_core_qualifiers#IsPartOf "/>

<rdf:value resource=" http://www.sendmail.org/antirelay.Parse0.txt "/>

</dc:Relation>

</rdf:Description>

Metadata and XML can bring disparate resources together and resolve information integration problems for the enterprise. Whether or not they can provide a Knowledge Base as an expert system for the helpdesk depends on further research and development in data mining by Oracle Corp. or DB2, or by third parties like Informatica and SAS. However, we are beginning to see the implications of data mining to generate new sales and increase e-business. In the meantime, metadata is being leveraged by many KM solutions aimed at integrating information in the enterprise.

Lotus Development Corp. and Semantix are positioning themselves as gateways to enterprise knowledge using agents. Lotus' Discovery Server uses agents or spiders to read or mine data repositories and filter content to normalize it. The normalization processes analyze available metadata along with content and provide context through a taxonomy built from the documents themselves. Using adaptive learning or data mining techniques, Discovery Server also distinguishes experts from knowledge workers by provenance. Tokens identify unique points of access to information and experts stored in a DB2 database.

Semantix's gateway approach to enterprise information makes use of a sophisticated search engine that brings a multiplicity of data and metadata from repositories and document management systems. For example, Semantix enables "adapters" that index repositories from LiveLink, Documentum, FileNet, Notes and MS-Exchange. Semantix brokers search requests from desktop integration points utilizing Microsoft search functions in MS-Word, Excel and PowerPoint through a query server.