Phase 1: Develop Methods and Tools

This project will begin with a "proof of concept" effort that will research and validate the methods and approaches that will give direction to the tasks that will follow. During this initial phase of the project the project team will research candidate knowledge representations appropriate for the nature and content of the historical artifacts. Of key interest here are the knowledge acquisition approaches that are best suited to extracting knowledge from domain experts who are most familiar with the artifacts. Subsequently, a set of automated Web-based tools will be prototyped and tested on a representative subset of the actual Freedmen’s Bureau records that are to be part of the larger effort. After the prototype tools have been exercised on this subset, there will be revisions made to the knowledge acquisition methods and refinements made to the tools that support these methods.

Phase 2: Build Knowledge Utility

As the initial versions of the tools are being developed, we will work toward acquiring additional Freedmen’s Bureau records along with historical artifacts from other archival projects underway at Howard University and elsewhere. After the directors of the other projects agree to submit their collections to the knowledge system, we will survey and analyze the needs of the various users of these collections to determine how AKADS can best serve them. In many cases, the artifacts will need to be digitized and the knowledge about these images represented using the approaches and tools developed in Phase I. The images and their related knowledge will be converted into "digital objects" and compiled into a structured knowledge repository.

Next, the project team will focus on the design and development of components for the document and knowledge retrieval agents. Based on the results of the survey of users' needs, we will determine and document user requirements. Using these requirements as a basis, we will formulate the search algorithms and inference engines that with help satisfy users' needs. Using the structured repository of digital objects and the algorithms for searching these knowledge structures, the "smart" search engine composed of intelligent agents will be designed. After the design of the knowledge/document retrieval agents is completed, a prototype set of software agents will be constructed (programmed) and tested. It is expected that negative test results will require some redesign of the search algorithm and inference engine components. Once the prototype agents are fully tested, the task of building Web services and an appropriate Web site can begin. The set of Web services, the knowledge repository and the user Web site will be the nucleus components of the fully functional and integrated AKADS system.

Phase 3: Build and Test Website

During this phase, the user requirements will be used to influence the design of the user Web pages and other interfaces for the AKADS Web site. The group of users who enumerated their needs in the previous phase will be asked to perform the acceptance testing for AKADS. It is anticipated that this systems-level testing will reveal necessary changes in the design of the user interfaces and the Web site. Once the group of potential system users and others on the project team deems that the AKADS knowledge utility has successfully passed the testing phase, the document/knowledge retrieval agents will be integrated into Web site. The document/knowledge retrieval agents, the Web site and the user interfaces will comprise the "Web services" an integral component for the Semantic Web.

Phase 4: Deploy and Support the AKADS System

In the final phase of this project, the team will advertise AKADS and make the system available to users from any Internet-accessible location. During system operation, software errors and bugs will surface and be addressed. User access and utilization will be measured and an evaluation of user satisfaction will be conducted. Also, during system operation, the tools developed in Phase 1 will continue to be used by project team members and others outside the University to further populate the knowledge repository with digital objects. After a significant operational period, a post- implementation review will be conducted to identify any system functions that have not met the user requirements for which they were designed. This review will undoubtedly result in some modification of AKADS and the related documentation.