lundez / javasymspell Goto Github PK

I don't get the whole SuggestionStage thing. Comments say it's to help speed things up and be more memory efficient, but looking at the code I can't see how would do anything but the opposite. Creating and populating the staging might be faster, but in the end it still has to go in the final data structure. So it's just an intermediate data object. I must be missing something.

P.S. While I am not sure, because I don't quite get it, by using a Trie I think staging becomes a moot device anyway. Would like to be sure before I rip it out.

In chunkarray file SymSpell.SymSpell.SuggestionStage.Node is needed !!
But no description is given where to find it.
Help wanted

Hi!

I am trying your java version of SymSpell.
It works really well, however I have noticed a very strange behaviour.

I am using the provided demo, with frequency_dictionary_en_500_000.txt as dictionary ( find it here on SymSpell page )

If I type as input "the", only this word, without spaces, the suggestions are the following:

Enter input:   
the   
Lookup suggestion: the 1 6801236995 
Lookup suggestion: he 1 588605295
Lookup suggestion: they 1 326802098
Lookup suggestion: them 1 178377546
Lookup suggestion: she 1 140368041
Lookup suggestion: then 1 108733059
Lookup suggestion: thy 1 11355619
Lookup suggestion: thee 1 7761536
Lookup suggestion: tho 1 4200344
Lookup suggestion: tie 1 1819888
LookupCompound: the

The first is correctly the word itself, by the way it has a distance of 1 from the input word... It should be 0!

This is an isolated case, as with other correct words, I'm correctly getting a 0 distance.
Do you have any idea why this happens?

I would also like to add a completions method -- return all the words that start with a given prefix. It would be cool if it could also handle spelling errors in the prefix, but one step at a time.

Would this be an acceptable pull request?

Any plan to update this release to latest SymSpell version?
The v6.4 has introduced important changes most of them related to the new WordSegmentation. A minor change to maxEditDistance to speed up lookup.

Thank you.

Would this library be a good choice for use in an Android App? Or is it too memory intensive?

And if yes, any guidance? One problem that occurs to me is that it has to be possible to load the dictionary from an android asset file. So a function that takes an InputStream instead of a file name might be necessary.

Hi,

I couldn't manage to access the correctedString and segmentedString fields of Word Segmentation, so I locally changed them to public fields and now it works.

Not sure if that's the correct way of doing it though?

Thanks for the port!

@Lundez, thank you for all the hard work ;)