Greek dictionary/ dictionaries in general

[dim-geo]

Hi,

I have been using heliboard keyboard for a couple of years now.
Due to its limited Greek vocabulary I have created my own, based on openoffice dictionary.

All words have frequencies:
https://github.com/dim-geo/greekdictionary

1. Maybe it would be better if tt9 could use heliboard dictionaries. (So you don't have to fiddle too much with dictionaries)
2. if 1 is a bad idea, maybe you can provide a mechanism for creating /importing custom dictionaries for languages. I know there is a mechanism for a few words, but I am thinking about whole language dictionaries. (heliboard supports that)
3. if 1 & 2 are out of the question, can I contribute my Greek dictionary to you?
4. can I combine your Greek dictionary with mine? (your Greek dictionary has some typos, but other than that, it's a good start)

[sspanak]

Yes, I can replace the current dictionary with the one from Heliboard, or combine both files. I have handy scripts to do it in 5 minutes, so don't worry about that.

Unfortunately, I don't speak and don't use Greek, so I can't remove or correct the misspelled words. If you can suggest corrections that would be great. If you have a bigger and tested dictionary, that would be even better. I would definitely listen to an advice from a native speaker. What would be the most optimal approach (considering that none of us can manually check 1 million words for spelling errors)?

[dim-geo]

Hi,

I recommend to use my dictionary as a base, since it's spellchecked and has frequencies.
It has ~828000 words, so it should be fine size-wise.

I will clean up your extra words (~65000) via some scriptable ideas and manually.

For example, check words without any stress "´". If this word already exists in my dictionary with stress then ignore that word. If the word doesn't exist, I have to manually check it's validity and add it properly stressed.

For words with stress:
If your word has 2 stresses and the same word exists with one stress, then add it. (It's a weird edge case and probably your word is valid)

If your word has 1 stress, check if my dictionary has the same word but stress on another letter. If yes, ignore that word.

Of course those simple ideas above are not perfect and I need to check what words they will discard. (In order to see if my ideas are valid)

From my look so far your extra words are places (for example village names) and latin/english names. There are misspellings as well and indeed some valid words that are missing like 'σούσι' (sushi).

In any case I would highly recommend to allow users to install their own dictionaries as well. (idea 2 of my first post)
Then I can test my own dictionaries and come back with the best one.

BTW, you should have just received my first 0,99 euros from playstore ;)

[sspanak]

I recommend to use my dictionary as a base, since it's spellchecked and has frequencies.
It has ~828000 words, so it should be fine size-wise.

I will clean up your extra words (~65000) via some scriptable ideas and manually.

For example, check words without any stress "´". If this word already exists in my dictionary with stress then ignore that word. If the word doesn't exist, I have to manually check it's validity and add it properly stressed.

For words with stress:
If your word has 2 stresses and the same word exists with one stress, then add it. (It's a weird edge case and probably your word is valid)

If your word has 1 stress, check if my dictionary has the same word but stress on another letter. If yes, ignore that word.

Of course those simple ideas above are not perfect and I need to check what words they will discard. (In order to see if my ideas are valid)

I sounds like a solid plan. Please, go ahead. I would be happy to upgrade Greek with the final result.

From my look so far your extra words are places (for example village names) and latin/english names

I believe it is nice and useful for the dictionary to contain proper names. I would say, let's keep them.

In any case I would highly recommend to allow users to install their own dictionaries as well. (idea 2 of my first post)
Then I can test my own dictionaries and come back with the best one.

You can not simply import just any list of words. There are technical constraints about the possible characters, the frequencies, the formatting and whatnot. Some particular words may cause other words to disappear. I mean, they will be in the dictionary, but it won't be possible to type them. In other cases, the words may cause typing lag, or may even crash the app.

I do not plan to allow such installations, sorry.

Anyway, the app is open source. If you have technical knowledge, you can build yourself an APK with your own dictionary, test and experiment as much as you want. There are no limitations to that.

BTW, you should have just received my first 0,99 euros from playstore ;)

That's cool, thanks! FYI, I can't tell who you are. The orders contain only date, time and the original currency. And, it is the way it should be, in my opinion.