Limit wall to user language

[kleinermann]

Hi there,
I'm still new to Diaspora, so i'm not sure if this has already been pointed out somewhere. When i subscribe to tags in my feed i get posts in all sorts of languages. It would be great to limit this to the languages i actually understand.... Are there any thoughts on this?

[Raven24]

While I get that what you propose could make it easier for our diverse community to follow people's posts all around the world, I think the technical side of this feature might bear some problems.

If we use a webservice, the language of a post has to be determined in a background job and might fail randomly due to unforseeable downtime of the service. Also with an external service, we'd have to send (a part of) the message somewhere which definitely is a privacy issue.

I just googled around a bit, and I found this ruby gem for detecting languages: https://github.com/peterc/whatlanguage (MIT License, last commit 6 months ago). But it might not detect every language and the detected language might not be correct (but I guess a certain probability of false-negatives will never be completely eliminated).

This certainly needs some more discussion, maybe even a collaboration with the author of the mentioned gem.
@kleinermann I think it would be best, if you could move this over to the mailing list

[ghost]

I also think this would be a great feature. And automatically detecting a post's language isn't strictly necessary. It could be marked by a user. I could imagine selecting which languages I want to have in my stream and choosing a default language for my posts. When composing a new post there might also be a way to select a language other than my default choice.

I realise this might be a big thing to implement but if Diaspora* really does kick off and more and more users from different countries start using it, it will become increasingly more difficult to keep one's stream "clean".

[harryhaller02]

Quite a lot of the people are multilingual. We have people who not only post in 2 or more languages, but also send messages in which the languages are mixed.
So the user profile is not going to be a hard and fast guide, at all!
The aspects were designed to cover this situation.
The other alternative is to use the "ignore" button on users who speak the wrong language.
Another possibilty would be to allow the user to decide to only receive messages from certain pods.

[ghost]

You said it yourself, some users are multilingual. So ignoring them is a horrible workaround. Maybe I only want to ignore the posts in language A, not the ones in language B.

[tssm]

You could specify in the settings which languages you can read.

[jimis]

I'm a new user to diaspora and I'm liking it so far, besides this language issue: for example I subscribe to #linux and my stream gets filled up with tons of posts that mean nothing to me.

I propose the following solution, I think it's pretty easy: every post should have a "language" property (independent of tags), by default set to "english". Optionally a user can change it to signify the language of his message, choosing from a defined set of languages.

On the UI it could work like that: in the edit box next to the insert-photo and geotag buttons, a small, faded out button labelled "en" should exist. The user can click on it to change the message's language property. The UI should automatically detect any non-ASCII characters in the post and automatically bold-ify this button and show a small bubble saying "change the language of this message".

[axaluss]

From #4764 :
It is annoying to see posts in languages you don't understand.
Therefore it would be nice to be able to filter out those posts in foreign languages.
For this a selection of understandable languages could be implemented as an option for each user.

The language detection presentation could be implemented automatic considering post and link content.
While posting the user is asked if the automatic detection of language xy for the post was correct and/or give a list of most probable languages.

The language filter could be implemented by asking the user which languages he likes to read.

The language detection could be implemented considering the n most frequent words in each language and countig their occurence in each post.
For example like that:
https://gist.github.com/axaluss/9345239

[h3r0n]

Please, add this feature in the same way axaluss said

[axaluss]

Here is a paper about language detection.
http://odur.let.rug.nl/~vannoord/TextCat/textcat.pdf
http://en.wikipedia.org/wiki/N-gram

[goobertron]

This is currently being worked on in #4998.

[deutrino]

This may be abandoned in #4998.

[davidbaumann]

I think, every user should define the languages he understands.
When posting, the post is marked with the language (multiple languages supported?).

Now the other users won't see this post, if they don't have the language configured...

[kp666]

how much have design goals changed here?

[Changaco]

I would really like to see this implemented, and I've added a $15 bounty for it.

I suggest doing this incrementally: forget about language autodetection for now, just ask the user in what language their new message is (the default should be whatever language the UI is being shown to them in), and which languages they want to see on their wall.

Autodetection can be added later to try to spot mislabeled messages.

[Mikaela]

Nice to see there was already GitHub issue, I commented this to Loomio:

I would like to have option to filter languages other than Finnish and English as I don't understand other languages.

Currently I can unaspect people who mainly write in languages I don't understand, but there are still followed hashtags which aren't restricted to one specific language as either the same word exists in both languages or everyone just uses the word instead of whatever the word for their language is.

[davidbaumann]

So I guess there are some points, that can be implemented independently

1. Users can select multiple languages they understand. The selected language for the interface will be added automatically.
   There is a selection for "Show all posts" "Show posts with no language tag and which I understand" "Show only posts with my language tag"
   Default will be "Show all posts"

2. When posting, there should be a small, grayed button with the language for the post "en" "de", and whatever, allowing to select between all language the user understands
   So the posts will have a language tag.

I think we should implement this.