How to Write a Spelling Corrector

December 11, 2007 – 10:49 pm
Posted in C++

I wrote a C++ version of the Norvig’s spelling corrector. The theory and other language implementations are in his website, and my version can be found at my github:

https://github.com/felipefarinon/spellingcorrector

19 Comments

- JAS
- Posted June 4, 2008 at 2:18 am
- Permalink
- Reply
HOLA… Q TAL…
Ps yo ando buscando opciones d como hacer un este proyecto pero poner agregarlo al openoffice con palabras de alguna lengua indigena…. y necesito material q me sirva de guía primero para entnder como se hace y dspués aterrizarlo al problema q quiero resolver…

Me pueds mandar x favor el código… O más o menos orientarme al respecto d q si es posible hacer esto que quiero… Desarrollar un corrector que tenga diccionarios de palabras en alguna lengua indigena y psoteriormente agregarlo al openoffice, o sólo se podrían agregar diccionarios???…

En vd necsito esta información y agradeceré me contest…

GRACIAS…
- Pablo
- Posted September 24, 2008 at 5:25 pm
- Permalink
- Reply
Para que coño quieres meter una lengua indigena si los tios estos no saben ni que es un ordenador !!
- - Golfizta
  - Posted March 28, 2011 at 7:42 am
  - Permalink
  - Reply
  Habra gente que quiera aprender a usar alguno 🙂
- Oleg
- Posted October 10, 2008 at 7:33 am
- Permalink
- Reply
Hi Felipe,
Well done, thank you.
How do you think is it possible to do some optimization to make code faster?

Best regards,
Oleg.
- scarvenger
- Posted October 10, 2008 at 1:14 pm
- Permalink
- Reply
Hey Oleg,

Thank You 🙂
The code is actually running pretty fast but yes, it can still be improved, I think I will write another post in the blog just for that subject ok?

Regards,
Felipe Farinon.
- Yassine Benjira
- Posted March 18, 2009 at 10:28 pm
- Permalink
- Reply
Thanks for the code
I think there’s a bug in the code. The transposition edit should be
word.substr(0, i) + word[i+1] + word[i] + word.substr(i + 2) instead of word.substr(0, i) + word[i+1] + word.substr(i + 2)
- - felipefarinon
  - Posted November 20, 2016 at 6:37 pm
  - Permalink
  - Reply
  Yes, there was, thank you for pointing it. The fixed code is at github.
- LORENA
- Posted April 9, 2010 at 11:03 pm
- Permalink
- Reply
podrias acesorarme para hacer un corrector ortografico en visual basic
- LORENA
- Posted April 9, 2010 at 11:05 pm
- Permalink
- Reply
podrias mandarme lo comentarios a mi correo o a cualquier persona que lea este mensaje lyvelasquez_09@yahoo.com
- Devanagari Fonts
- Posted December 13, 2011 at 8:47 am
- Permalink
- Reply
Awesome…thanks
- steve Jew
- Posted September 22, 2012 at 10:04 pm
- Permalink
- Reply
nice code. 🙂
- Edson
- Posted January 14, 2013 at 9:55 am
- Permalink
- Reply
Hi, i got a problem with this program, can you tell me witch tool do you use ?
and what is the type of file for the last program , and his complete name ?
Thank you,
- genti
- Posted July 9, 2013 at 10:32 am
- Permalink
- Reply
hey Felipe,
I would like to use ur code in a mex file on matlab. I am working on an OCR application and I want to implement ur code for post processing of words. The link for the word database is broken. Can you shoot me an update for that link on my email. Thanks in advance.
- iandun
- Posted July 30, 2013 at 2:45 pm
- Permalink
- Reply
I am using this in my spell checking library libspellcheck (libspellcheck.codeplex.com)
- Vivek Sharma
- Posted March 6, 2015 at 9:56 am
- Permalink
- Reply
I think you missed word[i] after word[i+1] in transposition section of edit(). As transposition assume to swap to adjacent. Please correct me if I am wrong. Great work though.
- - felipefarinon
  - Posted November 20, 2016 at 5:56 pm
  - Permalink
  - Reply
  You are right, thank you for spotting it. This made the corrector fail in two of the tests listed by Norvig, the ones which had the word ‘peotry’ and ‘peotryy’.
  
  I’m uploading the correction.
- Giulio
- Posted January 14, 2016 at 10:42 pm
- Permalink
- Reply
Hi Felipe, I took some parts of your code as well, thanks!
I think I found a little bug other than the one pointed by Yassine: in the load function the “++i” increment makes you skip the first letter of the first word of the dictionary file, leading you to record a wrong word.
Giulio
- - felipefarinon
  - Posted November 20, 2016 at 5:56 pm
  - Permalink
  - Reply
  Hi Giulio, you are right as well. I’m uploading the correction.
- arjun
- Posted July 11, 2017 at 4:09 pm
- Permalink
- Reply
You can write a spell corrector using elasticsearch. Here is the link https://arjunjs.wordpress.com/2017/04/12/elasticsearch-building-a-simple-spell-corrector-with-elasticsearch/ .