6 COMPARISON OF ACTUAL AND PREVIOUSLY PLANNED R&D ACTIVITIES

6 COMPARISON OF ACTUAL AND PREVIOUSLY PLANNED R&D ACTIVITIES

In this section, we compare our planned activities for the previous year, as laid out in the Co-operative Agreement No. IRI-9411330, with our actual activities. For convenience, we first state the planned activities as laid out in the Co-operative Agreement.

1. Previously Planned R&D Activities

The tasks that were laid out in the Co-operative Agreement for the period 03/01/95-02/29/96 included the general task of testing and deploying the RP and using interactions between users/providers and the system as a basis for a comprehensive study of the user requirements. We also planned to develop detailed specifications of each of the Alexandria testbed components and to commence initial work on their implementation.

Specifically planned tasks, as stated in the Co-operative Agreement, included:

1. [Task T1] The system integration team will test the rapid prototype and will evaluate the test results. They will also develop module test plans for each component of the main testbed system. They will develop a system security and privacy policy for the testbed based on the security requirements determined in period I. They will also determine the security and privacy policy needed for each testbed component.
2.  [Task T2] The information systems team will interact closely with the rapid prototype developers, and will use the rapid prototype to evaluate Arc/Info and capture usage scenarios and queries. In relation to the main testbed, the team will identify a commercial database system for the catalog component, adapt the CMS (computational modeling system) data model for maps, images and other spatially indexed materials, and integrate feature extraction and multidimensional indexing for image data. In addition, the team will evaluate and select off the shelf software and hardware for implementing the storage component.
3.  [Task T3] The user interface team will commence the design of a query language for the Alexandria data model and, in particular, the design of a visual query language (e.g., query templates, query-by-example). They will also study implementation issues for such languages. A visual shell including history, profile, and visual grep and parsing mechanisms will be designed. Browsing capability based on hierarchical wavelet image decompositions will be investigated in collaboration with the image processing team.
4.  [Task T4] The image processing team will continue with its investigation of different wavelet decompositions for image data, and select the most efficient for storage and browsing of image data. The team will also work on content-based retrieval and will interact with the information systems team to implement efficient indexing mechanisms for image data.
5.  [Task T5] The parallel processing team will use the CS Department's Meiko CS-2 parallel computer to simulate a network of workstations for testing issues. They will also develop specifications and algorithms for parallel processing at the level of the individual components of the Alexandria testbed.
6.  [Task T6] The library requirements team will gather digital data sets ( e.g. DEM's for the United States at 3 arc/sec and selected 1 arc/sec files; selected DOQ's, as available, for California; and selected DCW data layers for the U. S.) and load them into the rapid prototype. They will test metadata standards against a wide variety of spatial data sets and load catalog records for prototype testing. Both data and metadata will be tested using the "skeleton" interface of the ArcInfo prototype.
7.  [Task T7] The user evaluation team will study user evaluation of the rapid prototype system. The team will analyze the user evaluation data and provide an assessment for the Alexandria development teams.

6.2 Actual Versus Agreed R&D Activities

All of these tasks, and more, have been successfully completed. The manner in which they have been completed, however, differs from that of our expectations at the time of the writing of the proposal. The major cause for these differences is the important changes that have recently occurred, and continue to occur, in Internet-related technologies. It is clear that the only reasonable strategy at present for developing and Internet-based DL is to remain highly responsive and adaptive in the face of these rapidly-occurring changes.

The main difference between our original plans and our actual activities occurred on the completion of the RP. The idea of moving to a web-based system, while not in our original plans, became the obvious approach to take by late 1994. Hence the design and implementation of the initial WP by 11/01/95 was not in our original plans. This strategic approach has undoubtedly accelerated the rate of development of the testbed system.

As noted above, the WP system follows the basic four-component architecture initially envisaged (i.e. interface, catalog, storage, and ingest components). We note, however, that not only were the four architectural components of the WP designed in accordance with the agreement, but they were also implemented in terms of a working system, which was not part of the contractual agreement.

We now compare the planned and actual activities in greater detail. In terms of the various tasks, we note that for:

1.  [Task T1] The RP was tested and the test results were evaluated, with a good characterization of the ArcView II interface and the Sybase DBMS that underlay the catalog component. Since ArcView is not currently extensible to a web environment, some of the test results are not currently applicable. The Testbed Team is now testing the various components of the WP.
2.  [Task T2] The Interface Design and Evaluation Team has captured usage scenarios and queries. The Information Systems Team is currently interacting with the Testbed Team in evaluating four commercial DBMS: Sybase, Illustra, Oracle, O2. Illustra and Sybase as being used to support metadata. Hence a relational data model is being used for maps, images, and other spatially-indexed materials. Feature extraction and multidimensional indexing for image data has been developed and integrated into the WP. The CMS model, as well as a variety of knowledge representation schemes, are currently being evaluated in relation to acting as advanced data models for the catalog metadata. Several secondary and tertiary storage systems have been evaluated and the ADL is currently attempting to purchase a tertiary system.
3.  [Task T3] The UI of the WP incorporates our current design of a simple query language, and browsing based on hierarchical wavelet decompositions has not only been investigated, but also implemented in the WP.
4.  [Task T4] The Image Processing Team has made good progress on investigating efficient storage mechanisms for image date. They have also made major progress in the area of content-based retrieval using texture, which has been implemented in the WP. Indexing mechanisms have been designed and implemented for this information.
5.  [Task T5] The Performance and Parallel Processing Team has developed and implemented efficient procedures for networking workstations to act as WWW servers. They have also designed, implemented, and tested algorithms for parallel wavelet decompositions.
6.  [Task T6] The Library Team has loaded test datasets of many kinds as well as a metadatabase of over 0.5M records. They have tested and implemented various metadata standards, including FGDC and USMARC.
7.  [Task T7] The Interface Design and Evaluation Team has evaluated the interface of the RP using, in part, a CD of the RP that was distributed to 3000 users for evaluation. They have designed, applied, and analyzed evaluation procedures for the WP and used the information in designing new versions of the WP interface.

Finally we reiterate our current belief that the basic architecture of the WP provides an solid framework on which to base future developments. In particular, we believe that although the system that will exist in 08/31/98 will differ in major ways from the current WP, its evolution from the WP will occur by incremental modifications and extensions of this basic WP.