CENDI ANNUAL PLANNING MEETING

Welcome

Tom Pedtke began the meeting at 9:05 am. The Secretariat briefed the members on the logistics of the meeting and provided additions to the Planning Book. The Chair and Secretariat will take the contents of the discussions and draft a plan for approval at a later date.

Mr. Pedtke also introduced the concept of CENDI as an enterprise network. While all the CENDI agencies perform similar information management functions, each has its strengths and capabilities that can be used to benefit the whole. We interact with one another and often share customers.

As a means of expanding our view of an enterprise network and what it can be, MITRE was asked to present their view of an enterprise and to demonstrate current and future technologies that support this view.

KEYNOTE: MITRE 'S CORPORATE INTRANET, THE MII: STATE OF THE ART AND BEYOND (George Marling)

George Marling introduced the concept of enterprise thinking. Thinking and acting as an enterprise is based on structured organizational boundaries that are permeable. Information and ideas flow across boundaries. This allows the various organizations within the enterprise to relate and cooperate in new and different ways.

Mr. Marling then introduced the presentations dealing with the MITRE Information Infrastructure (MII) corporate Intranet and technologies being developed to enhance the knowledge management capabilities of that Intranet.

A Corporate Intranet: MITRE 's Information Infrastructure (Dolores Derrington, Project Manager, Center for Integrated Intelligence Systems, MITRE)

Dolores Derrington, project leader in the CIIS, described the environment that led to the development of the MII. MITRE has more than 4000 staff at numerous sites that are geographically dispersed including sites overseas. MITRE also has a heterogeneous computer environment, with a mixture of MACs, UNIX workstations, and PCs with Windows 95 or NT environments. MITRE acknowledges that it is easier to solve a problem if you have a homogeneous environment, but a homogeneous environment may not be the best for the functions that are to be performed. Prior to the introduction of the MITRE Intranet, MITRE had fundamental building blocks in place, including an existing Ethernet LAN, gateways, routers, and firewalls. MITRE also had the benefit of being a technology company accustomed to and driven by its customers to be at the front edge of technology.

The MII was a corporate-level initiative originally deployed in May 1995. The applications and information resources have continued to evolve since then. Virtually all desktops in MITRE are accessing the MII regularly. There are approximately 5,400 unique users and 4.3 million hits/month. More than 200 corporately managed collections of information resources and databases can be accessed.

The use of open and adopted standards for its architectural components makes it relatively easy to swap products. This is the case with Web browsers that enable access across different platforms. MITRE has a license for Netscape but, because the PCs are being sold with Internet Explorer, there are a growing number of Explorer users, often with multiple versions. MITRE has a 216-color, corporate-wide standard that reduces interface differences among the various Web browsers. The difference in Java across platforms requires MITRE to test its applications on these platforms to be sure platform dependencies are resolved.

The MII provides an infrastructure that allows people to locate appropriate expertise and resources. It supports collaboration by facilitating the sharing and locating of information that has been provided for the good of the enterprise. The ability to "know what it knows" or, more appropriately, what its people know, reduces redundant effort, leveraging MITRE 's intellectual capital for its customers. The MII also increases productivity by providing timely information on the user 's desktop. The use of this environment has reduced development costs and resulted in more timely delivery of products. It also serves as a testbed for products under sponsorship by its customers.

The basic capabilities of the MII include a "phonebook", a corporate handbook, drag and drop publishing, searching and profiling, navigation and browsing, integration across applications, a homepage database, and authentication of the users.

The development of the MII began with the administrative information. The time cards are done via a Web interface to an Oracle database from a SmartTalk application. There is a find feature to navigate the project structures and codes. MITRE employees can be found by organization, project, a best match for spelling, or via the phonebook. Information is integrated from various parts of the enterprise; the picture that is displayed comes from the badge office file and the human resources information comes from the personnel system.

Collections of links to resources for new employees and managers, or "views" of the information space based on specific needs, are available online. The latest versions of company forms and procedures, including performance evaluations, are supported.

The employees are increasingly interested in more technical information. Personal homepages are available to employees. MITRE believes that this gives employees a way to learn the technology and to know what it is like to be a Web user.

The publishing environment allows the user to create a document via a native word processor. The document is then dragged and dropped to the Publishing environment. This environment automatically converts the document to HTML and publishes the document on the Web in a particular organizational structure. However, at this time, double storage space is needed for the native format and the stored HTML. MITRE is looking at ways to publish HTML on the fly, eliminating the stored HTML version. Internet Assistant is used to convert to HTML. MITRE 's own software is used to extract the metadata. Templates assist in the capturing of the metadata.

The DTIC and NASA thesauri are used, with modifications to provide additional granularity as needed. PL-Web from Personal Library Software is used for searching and as a profiling tool. If the user has Web-enabled electronic mail, the document can be automatically e-mailed from the publishing environment.

The Publishing system is on an NT server. The in-process work is in a separate folder. Use of the system, particularly for in-process work, has grown to the point where they can 't keep enough disk space available. Version control is a problem as the system grows, since the current system does not have a check-in or check-out procedure.

A unique and heavily used feature of the MII is the alphabetical index, that allows for synonyms and different language to be used as the access point to the site. This index is maintained by hand. It was originally crafted by two information specialists. While MITRE has no quantitative results, the qualitative results indicate that the intellectual effort is worthwhile. The maintenance activity is not great, except when the homepage is reorganized, which has occurred three times in the last three and a half years.

MITRE is working on integrating the MII with the Library CLS system. This will allow users to search across the metadata from the MII and traditional catalog records from the library system.

Several agencies asked about the availability of the MITRE technology. It is not MITRE 's business to be a software firm. They develop the technology and then seek to find an outlet to make it commercial. The phone book approach has been documented. Elements that were developed by MITRE are the intellectual property of MITRE, but they can be shared with the government as long as it is not shared with another contractor. MITRE does, however, do systems engineering and integration support.

Ms. Derrington indicated the whole MII is instrumental to the way MITRE is now doing business. They have a lot of information about use and information flow. The entire system was engineered in detail with a lot of human factors and people participating in its design. A paper on how the system was engineered can be found on the Web at the MITRE.org site.

Beyond the MII (Dr. Mark Maybury, Director, Advanced Information Systems Center, MITRE)

The MII is an information management system. Knowledge management, dealing with complex cognitive activities, is the next step.

Dr. Maybury presented an information analysis scenario as an example of how a user would interact in the new environment. The hypothetical analyst was given a short time frame in which to find salient information about chemical warfare proliferation. Dr. Maybury showed an integrated analyst 's environment in which multimedia content is handled, information is extracted from multiple documents, machine translation is performed, if necessary, and the user collaborates with colleagues and experts.

He began with a series of articles found on the CNN newsfeed received nightly. MITRE developed software to determine where newsfeed stories stop and start. The TIPSTER-related software is used to extract information on people, organizations and locations. A new semantic template can be trained within a few hours to allow information to go beyond the currently known topics.

Dr. Maybury then did an Alta Vista search. The results were processed through three levels of text analysis against Alta Vista 's metadata. Each gave more precise or broader information. Based on extraction of frequently occurring terms in documents retrieved via the Alta Vista search, the user can promote or demote the term weights and reorganize the Alta Vista results to get more relevant ranking. An automatic summarization tool is run against a selected HTML site to pick out only the important parts on which the analyst should focus. Documents can also be clustered by topic based on a lead document identified for similarity. Graphs that indicate word relationships based on semantic and syntactic analysis can be created. By comparing graphs from two different documents, it is possible to perform automatic comparative analysis of the two documents, identifying overlapping sentences of interest.

MITRE 's strength is in text analysis. In support of Open Source Information System (OSIS) for the intelligence community, MITRE is collaborating with Carnegie Mellon for use of its speech recognition. To date, MITRE has no solution to the problems of video and images in low bandwidth. For purposes of the demonstration, the video was stored locally. The rest of the demonstration was live.

Another interesting development at MITRE is the Collaborative Virtual Workspace which integrates audio, video and text interaction. MITRE has discussed the commercial development of this product with software companies. However, these companies have indicated that the MITRE development is still too advanced for the general marketplace.

Met task (Multilingual-named Entity Task) is an additional stream within TIPSTER that handles lower density languages. The task is funded by several agencies, including National Security Administration (NSA) under the TIPSTER Initiative. While MITRE does not fund this directly, the NSA is funding MITRE to do internal research in automated mining of foreign language dictionaries.

Dr. Maybury continued with the information analyst 's scenario by showing its integration with outside resources. KEAN, MITRE 's premiere knowledge management tool, allows a collection of users to assess Web pages and annotate them. The individuals provide descriptive and statistical quality ratings. A server manages the knowledge assessment and provides a requester with a composite quality analysis for the site.

Videoteleconferencing was also discussed. Dr. Maybury indicated that this was difficult to incorporate because the international standards are not yet in place. In three to five years, he expects videoconferencing to be commonplace.

Ms. Carroll presented the accomplishments for the year. Contacts were made with 64 organizations, doubling the number from the previous year. Many of these organizations were involved in the CENDI Cataloging Workshop. There were contacts with state and local government entities, including a group with which there had been no contact in previous years. Approximately 130 CENDI agency staff members were involved in CENDI activities during this year. The general consensus from the CENDI members was that the information exchange, networking and educational opportunities were very beneficial to the individual agencies.

CENDI is presenting two sessions at the upcoming OSS '97 (Open Source Solutions) conference to be held on September 3 - 4, 1997, in Crystal City, VA.

The group reviewed the current CENDI goals and modified the wording.

1) Coordination and Leadership - provide coordination and leadership for information exchange on important STI issues.

2) Joint Projects - undertake joint projects to increase effectiveness and use of STI systems.

3) Education - promote an understanding of STI and STI management.

The group held a brainstorming session about issues of relevance under the goals identified above. The outline of the discussion is presented below.

Title 44 (raise STI issues in this context)

Increase in the volume of Web information

• Controlling chaos in this environment

Intranets - including new customer infrastructures that do not allow open access from corporate Intranets to the public Internet

Target intermediaries that control intranet content

• finding versus end product
• invite intermediaries to brief CENDI
• issue of information censorship

What are the future issues?

STI is a smaller force than in the past - need to look more broadly

• commercial
• other areas
• NSDI
• reinventing STI

Extend leadership into new areas

be more proactive in broader issues

be leaders in advocating STI issues to these groups

Architecture of the future

CENDI Enterprise development

Issues of cataloging records

• metadata
• indexing initiatives
• classifications
• metathesauri

Computer Security and Digital Signatures

Focus on Digital Libraries and DL communities

Knowbots: government sites are being indexed regularly

• how does this relate to GILS?

COTS --- GOTS ---- Development

• What is the balance of investment that should be sought?

Information Architecture of the Future

Forecasting Newsletters - Hot Topics

The group discussed possible activities in support of the objectives outlined above. However, it is necessary to fine-tune the list of objectives at the next meeting before finalizing and requesting proposals for specific activities.