4 ms·
> We search for longest common substring. That idea can be improved upon: "length of largest common substring" is not the measure to use. At the least, you wan
by Someone 18y ago
> We search for longest common substring.
That idea can be improved upon: "length of largest common substring" is not the measure to use. At the least, you want to use Levenshtein distance <http://en.wikipedia.org/wiki/Levenshtein_distance> http://en.wikipedia.org/wiki/Levenshtein_distance>, but even that can be tweaked to weigh common typing 'errors' lower (for example, colour is closer to color than to clour)
> Then when a person searchs for a term, we use a proximity measure and then search
> for all the terms with in a certain proximity
This is a form of <http://en.wikipedia.org/wiki/Query_expansion> http://en.wikipedia.org/wiki/Query_expansion>