Talk:WordNet/Archives/2016

Incorrect Definitions

Over the last five years, I have worked extensively with WordNet. Along the way I have encountered a huge number of definitions that are obviously wrong, e.g. the troy ounce as a unit of weight in the apothecary system, which it is not. I am not counting borderline cases, e.g. megabyte defined as a million bytes instead of 2 to the exponent 20, since both definitions could be considered valid (although it would have been nice to include the original definition). The point is, WordNet's error rate is extremely high. It's as if the Wikipedia allowed anyone to contribute, but only authorized 2 or 3 people to error check and make corrections. The increasing reliance on WordNet reminds me of the story of the large commercial jetliner that had to taxi back to the gate in order to reboot Windows.Zyxwv99 (talk) 17:06, 14 November 2011 (UTC)

Wikipedia notes that a troy ounce is closely related to the ounce in the Apothecarie's system, so in fact wordnet's rather short definition seems to be correct at the level of granularity being considered (although it could not doubt be improved). I think the rate of actual errors, as opposed to sub-optimal definitions is not particularly high: I have parsed every definition sentence to extract various information, and done the same for several other published dictionaries, and did not notice any clear difference in accuracy. Francis Bond (talk) 15:14, 9 January 2016 (UTC)