Unless you have been living in a cave with an unlisted b-board mailbox, you are well aware of the current, hottest, trend in the information industry, that toward database access via OPAC systems, or Knowledge Networking . A couple of years ago, industry cognoscenti foretold the coming of the network and the impending arrival of its long-awaited by-product, the "library without walls."
While indeed the walls are tumbling down across the country, some of the homeless are no doubt wanting for shelter from the numerous "little details" which torment the installation of these networks. These are the items typically dispatched with waves of hands when decision makers and sales representatives meet.
Fortunately for the many libraries which have not yet implement Knowledge Networks, enough real experience has been accumulated in recent months to provide some clues on what to think about when planning to install a network. This will allow many to benefit from the occasional misfortune of the brave infopioneers.
Strategic Issues vs. Anal
There's no question that it is more fun to waft in images of imaging than it is to wallow in disk drives. Nevertheless, the seemingly mundane issues determine the utility of a system in the 20th century, while establishing a solid foundation for the 21st. An ideal installation will sweat the details of the present, yet have sufficient awareness of the general direction in which the future is headed in order to grow in an intelligent way. Recent experience has revealed a number of obvious landmarks and hazards. While it is unlikely that a couple years of experience can expose all possible obstacles, if those listed here receive careful attention, the resulting system will be well equipped to take on the unknown.
Quite simply, there's never enough. On the surface, this may seem puzzling, as this ought to be one of the more easily quantifiable items on the Network shopping list. The problem lies primarily in indexing overhead. In the process of making the database navigable by the machine (as well as the user), the data must be "indexed," a term not used with true accuracy in the library sense. The process involves establishing logical links between selected terms in the database, and their destination information records. The greater the number of terms used, the greater the overhead, and the more disk space consumed. Some automation vendors' indexing schemes will grow a database as much as ten times, turning a Davidian database like Magazine Index of 400 megabytes into a Goliath 4 Gigabytes!
One must question the utility of a database which has been grown this large by over-indexing. While it is arguable that subject, author, and title indexing of these fields in the bibliographic citation is important, it is probably also true that linking the CODEN field to a full text is going a bit too far.
To Print, or Not to Print?
That is the question for the vast majority of public-access terminals, which at present have no companion printing devices. Now that we have the ability for our patrons to search and display a state-of-the-art information service, do we utilize number 2 pencils as the output device?
Then there is the question of full text and images. While the idea of dedicating laser printers to hundreds of public access catalog workstations is enviable, the current cost of the technology makes it less so. Printer servers may make the most sense over the long run. Thr thrift plan may entail small groups (2-4 on a single table) of temrinals tied to a dotmatrix printer with an automatic switch, with larger groups of terminals feeding to laser printer centers. Of course, one must then contend with sorting articles from multiple patrons and locations.
Academic libraries represent the lion's share of those which have installed Knowledge Networks to date. …