http://www.edcenter.sdsu.edu/presentations/sdscinternposter2001/csrc.html

Development, Implementation, and Evaluation of the Computational Science Resource Community

SDSC Student Intern Poster Presentation 2001

Chris Harper, EdCenter REU Student
Jeff Sale, EdCenter Computer Resource Specialist
Kris Stewart, EdCenter Director

Abstract

The explosion of information on the Web has spawned a growing need to provide a more integrated and personalized interface for academic and research professionals needing access to such a vast wealth of information. To meet this need, the Education Center on Computational Science and Engineering (the "EdCenter") and the computational science group at Boston University have partnered with the California State University Center for Distributed Learning (CDL) to initiate the development of the EOT-PACI Computational Science Resource Community (CSRC).

The CSRC is a web-based portal offering customized access to a repository of computational science resources, with a particular focus on PACI-related resources.  The CSRC is implemented using the Online Community Starter Kit (OCSK), a tool developed by the CDL using the Enhydra open source API to promote online community building around educational object economies.

This REU project has made significant contributions to the ongoing improvement of the CSRC’s design by

1) Implementing strategies for updating the repository’s content, and

2) Developing customized tools for tracking and analyzing visitor’s navigational behaviors.

Introduction

The EdCenter’s mission of fostering the use of computational science and high-performance computing in the undergraduate curriculum has led us to recognize the need for online community-oriented "gateways" or "portals" to academic- and research-based content on the web.  These gateways must be designed to nurture a sense of community within a specific academic discipline and must transcend physical and political boundaries while supporting multiple disciplines.

Main Goal of the CSRC Project

To address this need, the EdCenter has partnered with Boston University and the CSU Center for Distributed Learning to establish the Computational Science Resource Community (CSRC, see Figure 1).  The CSRC is a web-based portal offering customized access to a repository of computational science learning objects, with a particular focus on EOT-PACI.  The CSRC is built with the Online Community Starter Kit (OCSK), a collection of Java servlet class libraries developed by the CDL using the Enhydra Application Programmer's Interface.  The OCSK is designed to serve as the web-based "front end" to an Oracle database repository of computational science learning objects.  The Multimedia Educational Repository for Learning and Online Teaching, developed by the CDL (MERLOT, http://www.merlot.org/), is perhaps the most successful example of such a repository on the web.

Specific Goals of this REU Project

1. Populating the CSRC Database: Initial familiarity with NPACI/Alliance and related computational science resources in different scientific disciplines should be gained by assisting with the collection of materials to populate the CSRC repository.  The CSRC is designed so that additional resources are to be contributed on a membership basis (see Figure 2).

2. Evaluating search agents for semi-automated searching: Collecting online resources that are up-to-date is an ongoing job.  Identifying existing search agents capable of helping automate this process should provide added time to focus on other project goals.

3. Implementation of database-driven user tracking and analysis of user behaviors: Research into interface design methodologies should be performed to identify how these repositories are best designed in order to maximize their usefulness to their target user community.  As part of the formative evaluation process for the CSRC, user tracking should help reveal problems with site design and user navigation.  Programs will be developed using java and cgi scripts to perform custom queries and analyses, and generate reports of web server log files.

Methods

1) Installing and ‘Seeding’ the CSRC: Our initial focus was to seed the database with NPACI/Alliance resources and other resources related to computational science and high-performance computing in undergraduate education.  Each resource to be entered was characterized with up to 18 parameters including Name, URL, Description, Target Audience, Subject, Author, Institution, Technical Format, and Source Code Availability.  To locate resources we used a combination of researching NPACI/Alliance literature and searching the web using existing popular search engines, which we found to be adequate at best.

2) Evaluating software agents for semi-automated searching: We proceeded to evaluate search agent technology to assess whether any agents currently existed that might enhance the search process with semi-automated searching.  Our hope was to identify either a complete program or some partial source code we could integrate into our interface to combine the search and data entry tasks into a single automated process.  Using a customized rubric for evaluation (see Table 1 below), our goal was to identify several tools worthy of further evaluation

3) Implementation of database-driven user tracking and analysis of user behaviors: Java servlets and Oracle technologies were the tools of choice for customized web tracking and analysis.  A simple interface to a range of customized SQL queries within a web browser was created to provide convenient access to data on user navigational behaviors.  (See Figure 3)

Table 1: Search Agent Evaluation Rubric

The above rubric was developed to evaluate web search agent features in order to identify a potential tool to assist with locating computation science education objects.  We were looking for either a tool to use “out of the box”, or one which could be modified with relative ease to suit our purposes.

Results

The CSRC was successfully installed and initially populated with over 100 resources.  Each resource.  Visit the CSRC Web Page online at http://www.edcenter.sdsu.edu/csrc/.  The CSRC is scheduled to go officially ‘public’ two weeks before Fall Semester, 2001, when we will initiate a two phase announcement, first with an extensive email announcement, followed by registration with at least 10 of the major search engines and directories. 

We are near completion of the customized web-tracking interface in time to track and analyse the expected ensuing changes in web site traffic as a result of these announcements.  The supplementary web-tracking interface may be found at http://edcenter.sdsu.edu/~harper/

Automated search agent technology is currently not mature enough ‘off-the-shelf’ to offer help locating useful resources for semi-automated entry into the CSRC.  We did manage to identify three programs with enough features (see our evaluation rubric in Table 1 above) to be worthy of further evaluation: ASPSeek, Jobo, and Copernic.  Copernic offers an excellent interface with a wide selection of search results management and organization features, but most of these features are missing from the shareware version and no source code is available.  Jobo was developed primarily for web site downloading, but the availability of the java source made it a good candidate for further modification as a general search agent.  ASPSeek had the prohibitive problem of consuming an excessive amount of RAM, at the expense of other applications.

Discussion

There is no substitute for gaining familiarity with a partnership as large as the PACIs than to populate a database of resources representing the work of all of the institutions involved in the partnership.  Over 100 resources related to computational science were entered into the CSRC repository, each with up to 18 different characterizations to assist users of the repository in identify resources useful to them.  Members themselves will make additional contributions to the repository when the CSRC goes public in Fall 2001.

The expertise gained using Java and Oracle in online database interface design while developing a real-world application such as the web-tracking interface is something not typically acquired in an undergraduate curriculum.  The web-tracking interface will be put to use in less than two weeks to evaluate response to some targeted marketing in the academic community.

Acknowledgements

Special thanks goes to Dr. Mikhail Burstein, the EdCenter’s former Computer Resource Specialist (CRS), and Kirsten Barber, current CRS, who helped with mentoring and supervision of my efforts.

[EdCenter Home ] [Projects] [News] [Training]
[NPACI Showroom] [People] [Resources] [Directions]