Live data from Hacker News

A look at search engines with their own indexes (2021)

seirdy.one

11–20 of 28 posts

Re: A look at search engines with their own indexes (2021)

#11

Missed exa.ai! Embeddings-based search engine with its own index

How does an embeddings based search work? Without hallucinating bad links?

Not sure what they are doing but embeddings and hallucination are completely separable imo (you can have hallucination even without embedding-based retrieval). Likely you have an embedding for the query which is close to the embedding of the doc for some measure of similarity. That could be semantic similarity or even user behavior.

Re: A look at search engines with their own indexes (2021)

#13

A little tangential but does anyone know if there are any modern web directories? I'm wondering because it seems like due to the amount of spam on the web there needs to be more human curation as opposed to algrothims deciding what websites are valuable or not.

https://ooh.directory/ is one

Re: A look at search engines with their own indexes (2021)

#18

I have been somewhat impressed by Mojeek, but it does have two obvious flaws: 1) It not really good for localized search, it might be if you're local to the US or UK. 2) No !bangs. Coming from Ecosia I frequently just do !w !maps !yt because I know where I want the answer to come from For English language searches, it completely usable, but not quite as good as Bing or Google. I really wanted to try to use Mojeek as…

Local search and location-aware search is probably Google's biggest moat against smaller search engines. Bing does it passably, but it's aguably still pretty bad.

What's worse is that it's probably hard to ever get working well without the internet-scale profiling Google has access to.

Post reply on HN