Tip-of-the-tongue search
Tip-of-the-tongue search is looking for a specific item that the searcher has seen before but cannot name, so the query describes it from memory instead of giving its title. Such queries are long, vague and often partly wrong: they mention approximate details, the context in which the person met the item, and what it was not. It is a hard form of known-item search for keyword-based engines.
The phrase comes from psychology, where Brown and McNeill (1966) studied the feeling of nearly recalling a word. In information retrieval it names a known-item task: TREC has run a Tip-of-the-Tongue track since 2023, starting with movies and extending to general topics in 2025.
Example
“A late-90s film about a man whose whole life is secretly a TV show, not The Matrix” is a tip-of-the-tongue search for The Truman Show (1998). Nothing in the query matches the title, so the engine has to work from the plot description.
How it works in NEEDLE
NEEDLE’s Scholar vertical includes a tip-of-the-tongue query style: a hedged, half-remembered description of a paper with approximate quantities, no names the paper introduces and at most two content words from its title. A hit means the paper’s identifier is in the top 10. See the live NEEDLE results.
FAQ
Why is tip-of-the-tongue search hard?
Tip-of-the-tongue search is hard because the query shares few words with the item’s page and includes uncertainty, exclusions and false memories, which keyword matching cannot use.
How is tip-of-the-tongue search related to known-item search?
Tip-of-the-tongue search is a kind of known-item search: there is exactly one correct item, but the searcher describes it from memory instead of naming it.
Related terms
Written by Keenable.
Published
