Research Support

Preparation of the PD/A CRSP Continuation Plan 1996-2001 entailed a review of current aquaculture literature and discussions with many aquaculturists to determine research needs and constraints to aquaculture development in Lesser Developed Countries. In addition to limited knowledge of various aspects of production systems, lack of access to training and to information were found to restrict aquaculture development. In response to these needs, the Continuation Plan created research support as a separate building block of its proposed research activities. The following components, i.e., Information Management and Networking, Central Database, and Educational Development are the three branches of the CRSPs research support activities. The prescribed reporting requirements and timelines of the grant form the major part of the work plan of the Information Management and Networking Component which is, therefore, not reported here. The work plan for the Central Database focuses on the use of new technologies to enhance database performance, to provide wider and faster access to the database as well as to improve the userfriendliness of the database. Extension of research results will be greatly strengthened through the logistical and financial support available through the Education Development Component.

Investigations to be Conducted

Database Management
Human Capacity Development

Database Management

Collaborating Institutions

Oregon State University

John Bolte

Introduction

The PD/A CRSP Central Database is a centralized data storage and public query-retrieval system for PD/A CRSP aquaculture research in food-fish production. The database is open to data submission from all PD/A CRSP-funded aquaculture research projects, as well as other aquaculture research efforts with PD/A CRSP-compatible objectives and full compliance to the PD/A CRSP standardized methodology. The database was first built in 1985 and maintained by the PD/A CRSP Management Entity. In 1993 the database transferred to the University of Hawaii at Hilo (UH). The resignation of the database manager at UH necessitated the search for a new home of the database. A restricted Request for Proposals was issued. Oregon State University (OSU) won the award, and the database was transferred from UH to OSU in May of 1996. Over the period 1985-1996, the database migrated through a series of computer operating systems and database software. When received at OSU, the database contained data from 82 experiments which had been completed under Work Plans 1-7 from 1983 to 1995. The data were organized by experiment and data type into 720 separate files, maintained in dbf format using FoxPro database software. The following work plan describes the planned improvements for the database.

Objectives
Significance
Approach

Objectives

1) Develop and enhance the Central Database of the PD/A CRSP.

2) Provide improved standards and reporting capabilities for CRSP researchers.

3) Enhance the availability of the data via the World Wide Web.

4) Provide for online dynamic summary representations of the data on the Internet.

Significance

The PD/A CRSP Central Database is the largest database on warmwater aquaculture containing standardized data. It is a unique resource for aquaculturists worldwide, providing necessary information to construct and test models of warmwater pond systems. The database is open to the general public for data use, and may be searched by research site, calendar year, fish species, and fish production methods. Climatic, water quality, and fish performance data may be viewed in raw or summary forms and in graphical or tabular formats. The planned interface at a designated web site provides the best possible access to the database for a range of user groups, including researchers, educators, extension agents, and producers.

Approach

Database Housing: The database will be moved in its current form to the OSU Bioresource Engineering Department at Oregon State University, where it will "live" on a College of Engineering network server. These servers are appropriate to house datasets such as the PD/A CRSPs for several reasons, including 1) they are maintained outside of the CRSP program, ensuring a repository remains regardless of the future of the CRSP program, 2) they are centrally maintained with regularly scheduled backups, 3) they provide security and access control for the database, and 4) they are readily accessible from the Internet. The Department of Bioresource Engineering is committed to maintaining the housing of the database indefinitely (see Attachment from James Moore, Head, Bioresource Engineering).

Database Enhancements: The database in its current form provides basic data capabilities only. The availability of "metadata" describing individual datasets from the point of view of quality control, unusual conditions or circumstances encountered during the experiments, etc. has been identified as a priority area for database enhancement by the Technical Committee. We would develop metadata standards for the CRSP dataset in conjunction with the Technical Committee and implement those standards in terms of specifying reporting requirements and inclusion in the database. Other enhancements will be pursued in conjunction with feedback from the Technical Committee and the Management Entity, particularly in the areas of improving reporting and quality control.

Database Reporting: The database manager will be responsible for ensuring the timely receipt of datasets from CRSP researchers and the inclusion of the data in the database in a timely manner. The database manager will clarify any ambiguity, possible errors and incomplete data submissions with personnel at the field sites, and will respond to data corrections submitted by end users. Data templates will be provided to field personnel to accommodate entry of data described in CRSP work plans. Quality assurance procedures will be developed and enhanced to ensure a high level of quality of the data, based on standard measures (e.g. range checking, completeness, etc.). Electronic mail-based reporting will be provided, and Web-based reporting explored, to facilitate the ability of CRSP researchers to provide datasets as quickly and as simply as possible.

Distribution of Datasets: CRSP data is currently distributed on request in the form of diskettes, and this option will continue to be made available. However, the Internet provides capabilities for greatly facilitating the distribution of data via the World Wide Web and other technologies such as ALMANAC and WAIS. The Central Database can potentially provide the PD/A CRSP with a high-profile presence on the Internet and be used to more broadly disseminate the tangible results of the CRSP experiments. Internet-based access can also improve the timeliness of data dispersal, allowing essentially instantaneous recovery of specific datasets. Interactive technologies such as the Web now provide the capability for "live" viewing of time series plots, cross-site comparisons, summary statistics, etc. through a web browser. We see this as a critical direction for the Central Database to move towards, both to enhance the usefulness of the data and to showcase the efforts of the PD/A CRSP to a broader audience.

As mentioned previously, there are several Internet-based technologies which would be potentially useful for data interaction. We propose to focus primarily on utilizing the capabilities of the World Wide Web to perform three primary functions: 1) downloading of specific datasets, 2) "live" viewing of data and associated summary information and 3) presentation and distribution of related information, in particular the database "roadmap" describing the Central Database.

The first of these, downloading specific datasets, is relatively straightforward to accomplish. Current Web browsers (e.g. Mosaic, Netscape) provide high level access to FTP (File Transfer Protocol) services, a standardized download mechanism supported by virtually all network servers as well as Macintosh, DOS, Windows, and UNIX clients. The Web and FTP allow the transfer of specific files to remote sites, simply by specifying a filename link on HTML page accessible by a local Web server. Filenames can be explicitly specified, or generated in response to a query for specific information. From the developer's point of view (as opposed to a user's point of view), this requires several things. First, a Web server must be available to the developer. Second, a "script" able to process query request from the server must be written, capable of handling the query information and generating an appropriate dataset for file transfer. Finally, an FTP transfer must be invoked to physically transfer the file to the remote user. Communication between the server and the script is accomplished through a "Common Gateway Interface", a standardized mechanism supported by all Web servers. Scripts are generally written in a high-level string processing language such as PERL, or in a low-level language such a "C". Writing appropriate scripts and synchronizing server requests with script processing is the most complicated part of the task, and can be relatively easy or relatively complex, depending on the flexibility incorporated into the query request.

The second approach we would focus on involves developing "live" online documents permitting the query, presentation and downloading of both specific baseline datasets as well as summary data in tabular and graphical form. Current-generation Web browsers support the inclusion of "applets", either as stand-alone programs which can be executed over the Web, or embedded in HTML pages. The "Java" language, recently developed by Sun Microsystems, is the premier example of such an approach and is rapidly becoming the accepted standard for development of such applets. Java provides a C++-like language which connects interface elements on a Web page with server-based applications doing background data processing. As such, it provides an extremely flexible cross-platform tool for providing executable content on the Web. Such capabilities could greatly extend the ability of the Web to provide presentation of CRSP data, and would greatly enhance the types of analyses that could be accomplished online. In particular, doing cross-site comparisons of data, presenting summary datasets and corresponding statistics, and plotting datasets would be implemented and made available. Other useful applications (e.g. running models on selected datasets and presentation of the results) would be explored after discussion with the technical committee, and implemented as resources allow.

Restricting access to the PD/A CRSP dataset on the Internet has been discussed but left unresolved by the Technical Committee. We plan to continue these discussions until consensus is reached. Depending on the decision of the Technical Committee, appropriate access restrictions, if any, will be implemented.

Linkage with Other Datasets: Several other groups are developing aquaculture-related information sources. Of particular interest are AquaNIC, an Internet-based aquacultural information service maintained at Purdue University, and FISHBASE, a database maintained by ICLARM. We are currently linked with the AquaNIC service via the POND^© effort, and will expand this linkage to provide access to the CRSP database from the AquaNIC Web page. Additionally, linkages with the FISHBASE database will be explored, although this is more problematic since the types of data maintained in FISHBASE are of a different nature than that supported by the PD/A CRSP, and they currently have no Internet access points.

We will also pursue linkages with CIESIN, a centralized repository for environmental datasets, to increase accessibility of the datasets and enhance the visibility of the PD/A CRSP program.

Linkages with other PD/A CRSP Research: An important component of this work will be collaboration with PD/A CRSP researchers to 1) ensure that the database structure is consistent with their needs from reporting and analysis perspectives, 2) ensure that data is reported in a timely fashion, 3) develop data collection and reporting standards that are well-defined and appropriate, 4) develop quality assurance and metadata standards that ensure the usefulness of the datasets, and 5) solicit input regarding appropriate Web-based database analysis tools. This collaboration will be accomplished through regular communications and through presentations and interactions at the annual meeting.

Summary: The Oregon State University DAST is well positioned to maintain the Central Database and implement supporting Web-based technologies. We have used the database in the past to support our modeling work and are familiar with the overall structure as well as current deficiencies. For Internet access, we maintain a Netscape Web server on a Windows NT Advanced Network Server dedicated to operation of the Biosystems Analysis Group Laboratory which hosts the OSU DAST effort. The Web pages supporting POND^© documentation and distribution of related materials are maintained on this server. Because we administer this server directly, it is straightforward for us (from an administrative perspective) to establish the necessary permissions, modify Web pages, and incorporate CGI and Java support into the server. In addition to a fluent understanding of database operations and maintenance, we also have considerable programming experience to support implementation of Web applets and query processing. Additionally, we currently have experience in Web page development, as demonstrated by the current POND^© pages and pages for similar research activities.