If you’re a linguistic geek, or just like stoning out on how words work, check out WordNet. I was told of this site in an email discussion with a reader, it’s an ongoing academic research project based out of Princeton.
This site (it’s also available as downloadable software) is basically a database of interconnected word meanings. The site says: “WordNet® is an online lexical reference system whose design is inspired by current psycholinguistic theories of human lexical memory .” Er, in other words, it’s a neat way to see the various “senses” a particular word might have. The online site has a Word Search function. Type in any word, say….”search“…and you’ll see it has five senses as a noun, and four as a verb. You can then explore various aspects of the word’s senses, including synonyms, derivations, and – really cool – hypernyms: “Search is a kind of…” and hyponyms “..is a kind of search.”
I’ll admit, this was my first time stumbling across the terms hypernym and hyponym and actually understanding why they matter. So why does this matter to Search writ large? Because one way to think about improving Search is for an engine to drill down on a particular query based on what sense of the word the searcher intended. In other words, when you type “jaguar” into the query box, which sense did you mean – the cat, the car, the team, the software? If a search engine can create “senses” of words on the fly, it might be able to create smart responses to difficult and high-results queries (AltaVista and others do something similar with clustering, but this technology has not been blessed by everyone as relevant enough..including Google). Think of Google’s spell checker, but with “senses” of words, instead of spellings of words. “Did you mean the cat?….” etc. Now, I have no idea if this particular implementation would be useful to a search engine, it probably has all sorts of problems. But it’s interesting to think about nevertheless.
Thanks to Steve K for this pointer and the conversation that provoked it.
(An aside – my email is firstname.lastname@example.org. I welcome email if a particular thought of yours is more comfortable in that medium as opposed to the site’s comments area. I’ve learned a lot from such exchanges).
2 thoughts on “Find O’ The Day: WordNet”
Yes. WorldNet is a lot of fun to play with. And there are handy APIs for doing all kinds of interesting things with it.
It’s five years old but still relevant:
WordNet: An Electronic Lexical Database by Christiane Fellbaum