How We Got Where We Are: A Brief Chronology

The creation of machine-readable databases and computer-based services has always been predicated on the availability of appropriate hardware and software. During the first generation (roughly 1949 to the late 1950s), very little happened because the machines were slow, had relatively little storage capacity, and were extremely expensive. In addition, most programming was done at the machine level a tedious process. Second generation computers (from the late 1950s through the early 1960s) used transistors instead of vacuum tubes, which meant they were faster, more reliable, held more data, and could be afforded by institutions smaller than the federal government. To facilitate programming, a number of higher-level languages were developed during the early years. FORTRAN was designed primarily for scientific and engineering applications; ALGOL, the first of the so-called procedure-oriented languages, provided an internationally recognized structure for program documentation; LISP eventually proved valuable in studying artificial intelligence; and COMIT, the first language designed specifically for text processing, was used in computational linguistics and early studies in information retrieval. Higher-level languages greatly facilitated software development, because programs using them were shorter, easier to understand, and could be used on a variety of computers, unlike programs written at the machine level. It was also during this period, in 1958, that Hans Peter Luhn of IBM described the mechanized production of keyword indexes as well as an automated current-awareness service called SDI (Selective Dissemination of Information).published or submitted for publicatio