Automatically suggest similar proposals for Area 51 site suggestions

Question

The problem: many Area 51 suggestions are similar / duplicates of other proposals. The classic example of this week was Astrophotography which was a duplicate of (wait for it...) Astrophotography.

The current solution: rely on a group of volunteers to check out the new proposals and use a combination of the poor search feature and their memory to identify and close duplicates (which works as long as five volunteers (the number required to close) do this).

The proposed solution: just as the other sites provide similar questions as soon as you've entered the question title, so Area 51 could identify similar proposals. This would:

1. Save a lot of manual work (identifying and closing new proposals).
2. Be more friendly for new users (who may wonder why their great ideas are closed).
3. Not be a lot of work for developers (since it is a reuse of an existing feature).

What do you think?

Oh, and could someone please close Astrophotography? Done thanks.

Update: if you do this please also add a check on the words code, programming and any of the popular languages (C++, Java, etc). Anyone adding a description with these words gets the message, Is this in conflict with Stack Overflow? This will also prevent all the SO subsets...

Update 2: if the previous update is a specific instance, perhaps a more general option could be added: that of adding synonyms at time of creation. E.g. when creating a proposal about firearms you could also add synonyms like shooting.

Bounty: it seems like I need to offer bounty to get some momentum here. So the bounty will be offered to whoever (in order of preference):

1. Resolves this issue (i.e. from SO team).
2. Anyone who can improve to this proposal, either by refining it or by explaining why it should not be implemented.
3. Anyone who adds good testimonials to increase the chance that this will get implemented.

If it gets implemented, it will save a lot of work so please support if you agree.

Bounty awarded: to @David for his implemented solution. In a couple of weeks it would be good to check if this is sufficient.

New related question: Automatically search existing sites for Area 51 site suggestions

Answer 1

This feature has been implemented since shortly after Area 51 launched. Screenshot:

There are some improvements we could make (using the description to match, and not just the title; improving search), but for several of these examples it seems like users just ignored the existing proposals.

UPDATE

We now match prefixes, so "astro" will match "astrophotography". We also show a bit of the audience, so users will be able to see whether it actually matches or not since titles aren't always very descriptive. Here's what it looks like now:

Answer 2

update Substring search is now implemented, thus delivering more potential dupe-candidates.

---

One problem is the lack of substring searching. In theory I could e.g. propose Physicists without ever becoming aware of the Physics proposal (although I'm sure the former would be considered the too subjective version anyway).