ACM and ACM SIGART (the special interest group on AI) have been running network accessible services for almost one year. ACM runs ACM Network Services (ACM NS), providing information about ACM and Internet access to ACM Members. SIGART runs the SIGART Electronic Information Service (SIGART EIS), which provides AI-related information, news, and announcements to the Internet Community [Wel93]. Each have adopted the strategy to make as much information available through current Internet technology as rapidly as possible, to develop a long-term model of the DL, and to develop a plan for evolving the existing service into the long-term model.
The technical issues identified in each of these efforts are in three groups: Management, Access, and Services. Management issues are broken into information management and system management; Access issues into physical (network) access and information access; and service issues concern what services to offer. Of these, information access is the most relevant to this workshop.
Information access is the manner in which users can get to the information stored in a DL, and it is the retrieval of that information that we are specifically concerned with. In the case of ACM NS, this information will include, among other things, all articles in all ACM journals. SIGART is concentrating more on dissemination of information like conference announcements than publications, but the ACM global plan is part of SIGARTs future, including publications. Users who have access to this facility will clearly be swamped with information unless they are given the ability to pluck out only what interests them. While this may sound like simply a query mechanism, there must be more to it than what traditional library database systems provide. The query facility must contain knowledge to help the user find the desired information.
Another of the key issues in the design of an information access scheme for this type of DL is the rapidly changing state of the library. The card-catalog approach of ``being there when you need it'' will no longer suffice in the information age, people need to be actively kept up to date, by the library, of changes which interest them. The query facility must support two modes of operation: as a search engine and as a filter. When acting as a search engine it ranges over all the information in the DL, searching for whatever matches the user's query. When acting as a filter, it is applied only to new information as it is added to the DL - information passing through the filter is sent to the user.