fuzzy search and typos

[gabzim]

Is there any way to support typos/misspellings of a word? Like a hamilton distance search where anything with a small distance is also included? eg: if I misspell tetx instead text. Is this supported?

[ts-thomas]

When I'm finished with the 0.7.0 release, I will think about how this could be implemented.

[gustavopch]

That's a very important feature. Are there plans to implement it?

[acauquil]

@ts-thomas With a fuzzy search you will make every other search libraries obselete. Can you imagine the impact on society ? Every release would be showcased at the Las Vegas flexsearch conference, you could have your own TV show, who knows?

Please, we crave this feature.

[a-tonchev]

I also need this feature. In other libraries you could provide e.g. fuzzyness: 0.2

[thgh]

Could be interesting to mention this on the README. Why would you include fuzzy search libraries in the benchmarks if flexsearch does not even support it?

[Hebilicious]

I just refactored from fuse.js to flexsearch, before realising that flexsearch doesn't support fuzzy search, so close from glory 😅 @ts-thomas you mentioned you'd look into it after 0.7.0, are you still interested in supporting this?

[davidpaulsson]

I also very much crave fuzzy search. Any updates on this?

[kevinlul]

Why does the README even claim fuzzy search when it's not a thing?

[Abhi13027]

@ts-thomas The world needs a warrior, Rise from the ashes.

[sujato]

This is an essential feature for us, we'd love to use flexsearch but without fuzzy search we'll have to use something else.

[manuelportela]

Dear @ts-thomas,

I recommended your fantastic lib to a friend who just came back to me that this was a bad consulting. I used your library often in the past and did not even notice that fuzzy search was not part of it. What made me think that some kind of fuzzyness is covered is due to the encoder overview in the docs.

I think it would be fair to clearly state that fuzzy is currently not part of the deal - somewhere way up in the docs. I would also love to get your opinion about this topic. Do you think fuzzy search is something that may be combined with you r current approach?

Since today seams all about AI, LLM, Embeddings, I would also would like to hear your opinion about the approach of using vectors for similarity search.

Sorry, I'am curious about all these questions but I do not expect an answer. I am very thankful for all your work you put in here. 0.8 looks lovely!