|
ACS Home Related Links:
Advanced Campus Services |
Art Vandenberg Director, Advanced Campus Services University Plaza, Atlanta, Georgia 30303-3083 E-mail: avandenberg@gsu.edu Office: (404) 463-9685 Fax: (404) 651-4563 MEMORANDUM To: ATTN: AEG-PROGRESS REPORT ( AEG@sun.com ; 650-786-3451) MS/MPK 15-204 901 San Antonio Road Palo Alto, CA 94303 From: Art Vandenberg , Georgia State University Subject: AEG Progress Report Reference: AEG Award: Sun Grant Approval Number EDUD: 7824-010460-US Project: Directory Services for Communities of Interest Institution: Georgia State University Principal Researcher: Art Vandenberg, 404-463-9685 Date: June 12, 2002 (April 2002 quarterly report) (Attachments)1.) How are you using the SUN equipment that was donated to you? We continue to use the Netra X1 machines for testing of approaches to promoting semantic homogeneity and interoperability of LDAP directories. The Sun 250 has been deployed at Georgia Tech location (Dr. Sham Navathe College of Computing) with a database (Oracle) to provide supporting data structures for representing LDAP schemas, metadata, and directory related resources. We are running iPlanet LDAP, an instance of Novell eDir on Solaris, and using Metamerge , metadirectory services engine, to help integrate and synchronize the components. As a result of participation in National Science Foundation Middleware Initiative (NMI) Integration Testbed project (Sept 2001 - Aug 2004) (see Attachment A), we are expecting to add GRIDS technology components as a part of the project. We would like to make the Netra X1s available to GRID community, initially testing it with Georgia State Physics & Astronomy. 2.) What is the progress of your project? (50-100 words; mention any product/software developed as a result of donation.) Our project began June 2001. We have developed an improved interface to our semantic facilitator , which now receives results of our clustering algorithm and displays LDAP objectClasses in clustered regions. The semantic facilitator prototype now includes an interface to specify new objects and attributes, a mechanism to create these objects in LDAP, a process to extract LDAP schema information and apply SOM clustering algorithm, and the display of the resulting cluster maps to the interface. Improvements in the clustering algorithm are being explored, including experiments to validate computer algorithms against human experts' clustering of objects. We are developing genetic algorithms to "evolve" the best combination of SOM parameters that will approach the precision of human experts. We have submitted two National Science Foundation Grant proposals related to this work, have submitted a paper, and are beginning a second (see Attachment B). Additionally, Dr. Vijay Vaishnavi's graduate students are preparing PhD Dissertation and Master's Thesis on topics related to this work. 3.) Do you have any URL pointers for the project? We expect this by end of summer term, August 2002. 4.) Are there any press releases/success stories about your project and, if so, where are they? URL? See above. Certainly, the inclusion of Georgia State University as part of the SURA funded National Science Foundation Middleware Initiative (NMI) Integration Testbed Program (Attachment A), is partly due to the SUN Academic Equipment Grant providing resources. I think it fair to say that the SUN Academic Equipment Grant has been a factor in bringing our research team together - availability of resources is always a plus. 5.) How has the donation of the SUN equipment, as compared to any other vendor's equipment, contributed to the success of your project? The Sun equipment helps our project by making it possible for us to deploy a reference model environment that is devoted to our research work. Clearly, availability of hardware and software (iPlanet) resources is important. Importantly, the SUN equipment has made it possible for us to respond to other initiatives (such as NMI Integration Testbed Program, Internet2 Middleware videoconferencing middleware working group, and other directory collaboration projects.) 6.) If a lab, how many students have used the lab and in what period of time? There are 4 students working on the project at Georgia State - 1 PhD, 2 Master's, 1 Undergraduate, and 3 at Georgia Tech (1 PhD, 2 Master's) during Spring Term 2002. Additional information: We will be investigating the deployment of GRID computing components as well as Internet2 Middleware directory objects as part of the NMI Integration Testbed.
Attachment A - Current Award Georgia State University National Science Foundation Middleware Initiative (NMI) IntegrationTestbed Executive Summary Georgia State is committed to integrating middleware into our campus to advance productivity for educators and researchers. We concur with the premise of the NMI Integration Testbed Program: "Successful middleware integration will allow research scientists and educators to rely on consistent, easy-to-use, non-intrusive and privacy-oriented services and interfaces to accomplish scientific collaboration and information sharing with great efficiency." Since September 1999 Georgia State University has worked actively to implement such a middleware infrastructure for Georgia State. In February 2000, the CIO established an Advanced Campus Services unit with a specific charge to deploy middleware architecture consistent with the national middleware initiatives of Internet2 and Educause. Georgia State University has provided leadership in the University System of Georgia for middleware activity and is actively participating in Internet2 Middleware activities. Summary of Georgia State Qualifications · SURA institutional membership · Research & education projects for videoconferencing, grid technology, and collaborative tools · Ability to test specific components of NMI-GRIDS and NMI-EDIT · Enterprise-wide identifiers, directory services, and objectClasses being deployed · A centralized authentication that can be leveraged toward centralized infrastructure consistent with NMI objectives · Significant institutional awareness of middleware; existing collaborations between IT and research faculty · Well qualified administrative and technical leads · Institutional commitment to state of the art middleware · Demonstrable capacity and willingness to focus resources · Commitment of Associate Provost/CIO J. Reid Christenberry to middleware infrastructure deployment Georgia State has research and education initiatives that can benefit from the integration of middleware, in the area of videoconferencing infrastructure, control of remote instrumentation and sharing of large (terabyte) datasets: Dr. Richard J. Welke - Global eManagement Consortium program:
Dr. William H. Nelson - Physics & Astronomy:
Dr. Vijay K. Vaishnavi, Dr. Shamkant Navathe (Georgia Tech), Dr. Carl Stucke - Collaborative Research: · Georgia State has established collaboration with faculty at Georgia State and Georgia Tech in the area of directories, metadirectory services. This collaboration brings together expertise in the areas of B2B and eCommerce (Dr. Vijay K. Vaishnavi), databases, ontology, distributed architectures (Dr. Shamkant Navathe), video collaboration (Mary Trauner), security (Navathe, Dr. Carl Stucke, Vaishnavi), and middleware architecture (Art Vandenberg). The research that is being investigated is closely related to the issues that NMI raises, such as how objectClass and attribute homogeneity can be promoted, resource discovery in a federated environment, and directories deployed for communities of interest. The NMI Integration Testbed presents a synergistic activity that can benefit, and gain benefit from, these existing activities. Georgia State is willing to embody the current "state" of middleware (as defined by NMI releases) through active and ongoing implementation of applicable NMI standards, processes, and components.
Attachment B - Pending Awards NSF Grant ProposalsNational Science Foundation, Information Technology Research Proposal No. 0220321 PI: Vijay K. Vaishnavi, PhD ITR-Enabling Virtual Communities: Proactively Promoting Semantic HomogeneityEnabling virtual communities brings a number of research issues to the fore, including discoverability, resolution of semantic heterogeneity, and use of standard structures and formats for interoperability . Individual web pages provide extremely flexible mechanisms for the description of any individual's particular interests, but they do not provide structured descriptions. Further, free form descriptions enable only keyword searches; this limits the precision of searches [3] and the automated processing of information [4]. Lightweight Directory Access Protocol (LDAP) directories overcome some of these limitations by enabling individuals to describe themselves in a structured manner; however, they simultaneously limit the flexibility of description. Specifically, while the information associated with any individual can be well-defined by using pre-defined objectClass tags, this same description may be quite limited due to the constraints of using only pre-defined tags. Indeed the very process of getting consensus on the definition of tags can be a constraining factor. This project proposes to develop, implement, and validate a semantic facilitator model that will provide appropriate mechanisms for:
Our proposal aims to extend LDAP so that it can be used not only to provide structured descriptions of individuals (as it already can), but also to enable the extension of description mechanisms so that they are semantically interoperable. Specifically, structured description mechanisms such as XML already exist, but these standards have not addressed the core challenge of determining shared meanings for those descriptive tags. LDAP provides a mechanism for the description of resources. This project takes the view that resources consist not only of data, but also metadata (descriptive tags.) The project aims to enable extensible description mechanisms by providing an environment in which metadata can be dynamically defined and proactively promoting the reuse of existing metadata (i.e. promoting semantic homogeneity). Both Kohonen Self-Organized Map (SOM) [33] and Latent Semantic Indexing (LSI) [34] have been used to do information indexing and retrieval, and have been shown to improve information indexing and retrieval speed and accuracy [35, 36, 37, 38]. This research will apply these approaches, among others, to the discovery and correct reuse of metadata . The project will begin with the case of attributes describing persons and their activities, such as might be used to extend the inetOrgPerson objectClass (cf. eduPerson ). A SOM-based user interface will be designed and implemented so that individuals can " provide flexible, complete descriptions of themselves. " The interface will use SOM and LSI to discover and present attributes to the user in a semantically meaningful way. The interface will help the user to map custom attributes into the domain of existing attributes - and so establish additional semantic meaning. SOM and LSI will be investigated as the means of processing the attribute space and discovering structural semantics, with particular focus on how these and other clustering algorithms may be adapted or optimized to work effectively in the LDAP directory space. National Science Foundation, ANI - Strategic Technology for the Internet Grant Proposal No. 0230900 PI: Mr. Art Vandenberg, MS Resource Discovery of Middleware Objects and Metadata Using Peer-to-Peer Components The World Wide Web provides an environment that abounds with rich information sources. Businesses, organizations, government and education use these Internet resources for communication and collaboration, facilitating dynamic interactions of virtual organizations . The discovery of resources and services is a core element of current research in middleware: "The layer of software between the network and the applications [that] provides services such as identification, authentication, authorization, directories, and security. By promoting standardization and interoperability, middleware will make advanced network applications much easier to use" [1]. Middleware supports not only physical organizations, but also virtual organizations that reach beyond enterprise boundaries, requiring methods for resource discovery in a distributed, inter-organizational environment. The dynamic richness of web resources pushes the bounds of middleware even toward self-aware, reflective middleware [2, 3, 4] that can evolve and adapt to dynamic conditions. Virtual organizations vary in their degree of permanence. For example, the GRID (virtual) community seeks to define its rules for direct access to computers and other resources across the Internet, and to enable collaboration in ways that transcend the boundaries of fixed organizational structure and leverage the resources of many. Foster et al. [5] observe, "The real and specific problem that underlies the Grid concept is coordinated resource sharing and problem solving in dynamic, multi-institutional virtual organizations." Due to the large-scale collaborations that the GRID community is supporting, it has a permanence that warrants the definition of standardized rules for resource description and interoperability. In contrast to GRID, the challenge of resource discovery for other virtual communities is that they are ad hoc and hence less permanent, so common definitions and standards are more difficult to achieve. Common ground is at best defined across only a minimum core of objects, and variations to the core may proliferate faster than the ability of ad hoc groups to come to consensus. This leads to the general research question addressed by the current research: How can resource description and discovery on the World Wide Web be improved for ad hoc virtual communities? In this research, we will focus on the resource description and discovery requirements of a video development community-of-interest (Video_CoI ), which is representative of other ad hoc virtual communities. Given the bandwidth improvements of Internet2, Video_CoI represents a significant community of individuals who wish to share not only specifications for access to local video equipment, but also technical knowledge about installation and operation of a video-conferencing infrastructure. Currently these information sharing capabilities are being addressed by the UCAID funded "Commons Portal" [6], a portal for video conferencing resources that enables videoconferencing persons at various levels (beginners, intermediate, expert.) to find answers to "how to..." questions. The objective of the proposed research is to develop a P2P resource discovery model and demonstrate how it can improve the resource discovery of an ad hoc virtual community such as Video_CoI . This will be done by developing a tool based on the P2P resource discovery model, applying it to the "Commons Portal" and demonstrating the improved resource discovery capabilities of the portal. Specifically, the approach will be to implement a tool to let a user: · contribute resource objects and metadata · extend metadata with "expert translations" that helps a community interoperate and potentially converge toward common objects · declare and retain control over their own expertise information · manage collaborations and inclusion in "assemblies of experts" The research approach has the characteristics of being simple and scalable, like the web itself, and will provide a mechanism to expose the resources available to a community by using peer-to-peer technology to leverage the expertise of its members, exposing the information known to a community and improving the quantity and quality of resource knowledge. The proposed research is expected to have a wide range of applications. The approach has the potential of speeding up the formation and sharing of results among communities of interest or among members of projects collaborating across the Internet, and encouraging the emergence of community-based shared semantics. The research team brings together academic researchers, IT practitioners, graduate and undergraduate students from two universities to develop a novel solution to a problem of wide interest and application. Working PaperD. Kuechler, V. Vaishnavi, and A. Vandenberg, "An Architecture to Support Communities of Interest Using Directory Services Capabilities," Computer Information Systems Working Paper, Georgia State University, 2002; submitted for Hawaii International Conference on System Sciences, 2003.
|
Last Updated: March 2, 2006