[userbinator]

fuzzy search

I do NOT want search to become any fuzzier than it already is.

See the great decline of Google's search results, which often don't even have all the words you're asking about and likely omits the one that's most important, for a great example.

[arkh]

> fuzzy search

> I do NOT want search to become any fuzzier than it already is.

For a specialized shop site you may want it. Search term: "something 150", the client is looking for a 1.5m something, if you're doing an exact text search your search engine will give you a lot of noise. Or you'll have to fiddle with synonyms, dictionaries and how you index your products with a huge chance to break other types of search queries.

[soco]

How many sites will have useful results to return for a "something 150"? Muzzle width? Bees? T-shirt size? Walking distance? You surely cannot want _all_ these categories yet you'll get them all in a list. I might be biased but today's fuzzy search is a dumpster fire, sites hating to return only two results so they bury anything relevant in a tidal wave of unrelated garbage. I have office mates like that and everybody hates them as well.

[arkh]

My current case is: whatever you'll look for in a hardware store. So anything yeah: muzzle width, wood length, protective gear, liquid quantities, animal food etc.

And depending on the client vertical they tend to not use the same vocabulary when looking for products.

But contrary to some other comments I know LLM are not magical tools and anything we use will require data to fine tune whatever base model we choose. And it will be used on top of standard text search not as a full replacement. I'm sure many companies are currently doing the exact same thing or will be soon enough.

[ZeroTalent]

But this is why LLMs are so amazing. They understand context and nuance, and they have reasoning skills now. So you will not get a long list of garbage from a good model.

[soco]

Do you know such models or is this wishful thinking?

[ZeroTalent]

o3, reasoner.com, and complex setups of "thinking" workflows for sonnet 3.7, gemini 2.5 pro, and o1-pro

Gemini 2.5 pro is basically free.

Also watsonx, but that's b2b.

[player1234]

Sounds like a trillion dollar killer app.

[kolinko]

Just because Google is doing it bad doesn’t mean it has to be bad.

[mrob]

I want both fuzzy search and exact search. Google still has the "I'm feeling lucky" button, so it can support multiple search buttons. It could default to fuzzy search and have an "I'm feeling unlucky" button for exact search.

[Joker_vD]

Just yesterday saw this great example: [0].

[0] https://grumpy.website/1642

[squiggleblaz]

I don't necessarily want search to become any fuzzier than it already is either, but what's happened has happened and I've already responded to the decline of traditional search engines. Nowadays I pretty much only search duckduckgo with site:(something), or else I ask perplexity the question and for some links. Traditional search engines now just give a thousand SEOed-to-death articles, probably generated by ai, from hundreds of pointless third party websites that just have the same basic milk.

It might be that it's worth it to bifurcate soon. Search indexes and AI engines, doing different roles. The index would have to be sorted with AI though - to focus on original and first-party material and to downrank ad-driving slop.