ACS Home

Related Links:
AIS World Internet2 NSF Dublin Core eduOrg

Advanced Campus Services
Information Systems & Technology
Georgia State University
P. O. Box 3968
Atlanta, Georgia 30302-3968
Phone +1 404 463 9685
Email: avandenberg@gsu.edu

ITR-Promoting Semantic Interoperability of Metadata for Directories of the Future

Directories provide a well-defined, general mechanism for describing enterprise resources within and enabling their discovery by individuals and applications.  LDAP (Lightweight Directory Access Protocol) directory services enable data sharing by defining both the metadata (schema information) and an access protocol.  Organizations in research, education, government, and business rely on LDAP services to represent in directories internal information about their people, services, and technology, and to provide identification, authentication and authorization to support data and application security.  Both the Internet2 higher education research community and the GRID community use LDAP structures describing people and computing resources to support their work.  While such directory enabled services can be provided within an organization, offering or accessing these services beyond the home organization requires significant coordination of directory architectures and standards for directory metadata.

The Internet2 and GRID research community appreciates the importance of interoperation supported by standard LDAP schema structures, and is devoting resources to definition of metadata, recommendations for best practices, and guidelines for directories.  However, there remain significant limitations to this approach:

• Interoperation depends on defining and agreeing to standard metadata.  Such standards development can be a lengthy process of consensus development and may not guarantee a complete solution.
• In the absence of complete agreement on standard metadata, heterogeneous solutions continue to present semantic issues (e.g., synonym, homonym problems) to be resolved, and reuse of metadata remains limited.
• Proactive, automated tools to promote semantic homogeneity of directory metadata are lacking.  Administrators create new objects, rather then reuse components, duplicating effort and introducing even more heterogeneity.
• LDAP directories deployed at thousands of higher education sites present a daunting volume and complexity of metadata for vendor specific structures.  Automated mechanisms to assist directory administrators in managing such volume and complexity across vendor specific implementations are lacking.

The proposed research will investigate an alternative to current approaches in directory metadata practice and will be based on the proposition that monitoring, clustering, and appropriate visualization of cross-organizational metadata can help to identify patterns of practice and evolving standards, and to promote and facilitate the reuse of such metadata, which in turn can lead to automatic dynamic evolution of standards.  The approach presents a number of issues, which will be the focus of the research.  The research builds on prior and ongoing work of the proposers in designing the Semantically Enhanced Enterprise Directory Services (SEEDS) architecture to promote semantic interoperability and in building the preliminary version of a research prototype tool based on the architecture.  Among the features of the architecture is an ability to continually monitor cross-organizational directory metadata and use it for identification and promotion of a baseline standard for metadata.  The proposed research will, in particular, focus on finding how best the cross-organizational directory metadata can be abstracted and visualized to maximize its reuse, centering on the design of the Semantic Facilitator TM (SM) component of the SEEDS architecture.  Prototyping of the SEEDS tool will be used as a vehicle for carrying out research, demonstrating feasibility of solutions, and experimentally validating the solutions.

The following are the main focus areas that will drive the research:

• To promote and facilitate the reuse of large amount of directory metadata requires an ability to automatically present the metadata at varying levels of abstraction.  Techniques for automated clustering of metadata at a level that matches human expertise, employing such techniques as Self-Organizing Maps and Latent Semantic Analysis/Latent Semantic Indexing, optimized using genetic algorithms, will be developed and the solutions experimentally validated.