I think it would be handy. I just searched for something using the word “Five”, and didn’t get the result I was looking for because it only had “5” on the page.
This could probably be achieved with custom implementation, eg at index time by adding the alternative string to a new metadata field. Or at query time by performing an additional search with alternative string. But it seems like a helpful feature to be in the core.
ahh nice one. Currently, Funnelback uses statistical based ranking algorithms in order to determine relevance (i.e. how many times a word appears). Going forward, we want to supplement this by investigating Natural Language Processing (NLP) in order to “understand” the user’s query in order to derive meaning and context.
In the meantime, you could use Funnelback’s synonyms system in order to expand queries such as one to include both one and 1.
e.g.
Oh yes, that’s a much better idea, and easy to manage!
There could be some limitations on managing this logic manually or even attempting something similar at index time, though:
Consider:
“345,678” ~ “three hundred and forty five thousand, six hundred and seventy-eight”
Do you want the page containing this number to be returned when searching for ‘three’, ‘five’, ‘six’ or ‘eight’?
A good question, but I suppose it would be the same as if you were searching for ‘3’, ‘5’, ‘6’, or ‘8’.
I think in practice you’re going to be searching for a number along with other word(s) (I was when I thought of this). Searching for the number “five” alone is not necessarily going to be a common search.
I think supporting this for single digits makes sense and dove tails quite nicely with the guidance given by the Commonwealth style manual:
Write out the number if it is a single-digit number (1-9) where there are no units or where the units are nonstandard (e.g., nine animals, five cells, or six-well plates).
