Re:Sheldon: The Erlang Spell Checker
by nudpiedo on Thursday November 25, 2021 @09:09AM
(#29341474)
The example was hilarious to read, and definitely a great feat to implement such a complex algorithm. I wonder why was Erlang picked and whether it will be offered as a service.
Also how does it know how to replace "correct" by "misspelled"? is it maybe a list of synonims with stadistic correlation based on the semantic domain?
Re:Sheldon: The Erlang Spell Checker
by vkatsuba on Saturday November 27, 2021 @06:20AM
(#29359332)
Not sure about the service in the current implementation, but it could be :-).
> I wonder why was Erlang picked and whether it will be offered as a service.
You see in open source exist a lot of spell checkers based on different languages - but in Erlang this is one of the first. In Erlang world already implemented a lot of interesting projects like messengers, chats, writted a lot of documentation where spell checking is need and important. Plus we already create some useful plugin for checking strings, binaries and comments in Erlang projects based on AST - but we will outline this in more detail in the next article, which I hope will be released early next week.
> Also how does it know how to replace "correct" by "misspelled"?
This is future implementation what we will try add into this library - so far this library offers only candidates for replacement. But we are gradually moving in this direction and will be collected more algorithms for it.
Re:Sheldon: The Erlang Spell Checker
by vkatsuba on Saturday November 27, 2021 @06:04AM
(#29359247)
The bazinga was added only for fun to keep the spirit of the famous TV series. We have already received some feedback that this is superfluous and not all users would like to see bazinga. For this reason, we have created a ticket and will work on it in the near future so that this behavior can be disabled or enabled through the configuration parameter.
Also how does it know how to replace "correct" by "misspelled"? is it maybe a list of synonims with stadistic correlation based on the semantic domain?
> I wonder why was Erlang picked and whether it will be offered as a service.
You see in open source exist a lot of spell checkers based on different languages - but in Erlang this is one of the first. In Erlang world already implemented a lot of interesting projects like messengers, chats, writted a lot of documentation where spell checking is need and important. Plus we already create some useful plugin for checking strings, binaries and comments in Erlang projects based on AST - but we will outline this in more detail in the next article, which I hope will be released early next week.
> Also how does it know how to replace "correct" by "misspelled"?
This is future implementation what we will try add into this library - so far this library offers only candidates for replacement. But we are gradually moving in this direction and will be collected more algorithms for it.
Thank you for sharing :)