Karen Spärck Jones showed that a term's retrieval value can be estimated from how rarely it appears across a collection, establishing the idea now known as inverse document frequency.
Gerard Salton's SMART research and the vector space model turned documents and queries into weighted term vectors, making graded similarity and ranked retrieval central to search.
Vannevar Bush's 1945 Memex proposal imagined a personal information machine built around associative trails rather than rigid indexes, anticipating central ideas in hypertext and knowledge navigation.