Tatoeba / tatoeba2

Tatoeba is a platform whose purpose is to create a collaborative and open dataset of sentences and their translations.
https://tatoeba.org
GNU Affero General Public License v3.0
707 stars 132 forks source link

Possibility to have different link types #1902

Open soliloquist-tatoeba opened 5 years ago

soliloquist-tatoeba commented 5 years ago

https://tatoeba.org/eng/wall/show_message/31999

Linking sentences with similar patterns (localized names) has been discussed on the thread above. Having more than one type of links might help to overcome this issue.

I'm copying my thoughts here.

Some users are in favor of using links for translations only, but in a linguistic sense, the definition of link is not restricted to translations. Synonymous sentences are semantically linked, too. Sentences with similar patterns are also logically linked. What could be linked and what couldn't is a bit relative and vague.

I think it would be more comfortable having different types of links: one for translations, one for synonymous or closely-related sentences in the same language, and one for sentences with similar patterns (or localized sentences like #7947998 ) in the same language. I'm not sure how all these categories could be shown at the same time without creating a visual chaos though. Maybe some filtering/hiding options or a tree view with collapsing/expanding feature might be necessary.

Here is a rough image of the idea.

https://prnt.sc/ny3d99

link

jiru commented 5 years ago

Thanks, that’s a good idea but I think it needs more refinement. We’d better continue the discussion on the Wall so that other Tatoeba members can join the discussion too. I’m assigning the unclear label for now. I encourage you to come back here once you come up with something more precise and have clearer answers to the questions in the description of the unclear label.

ckjpn commented 5 years ago

I think this would likely be confusing, especially if they are called links.

"Related" could include too many sentences -- opposite meanings, different tenses, sentences used in a similar situation, ...

Perhaps having different types of comments could accomplish what soliloquist-tatoeba wants.

https://github.com/Tatoeba/tatoeba2/issues/1830

I'm not sure I understand "Similar Pattern." If you mean "same pattern", this list could be very long. "I like ___." for example http://tatoeba.org/sentences/search?from=eng&trans_to=und&query=%22%5EI+like%22&to=tur&sort=words

soliloquist-tatoeba commented 5 years ago

By 'related', I meant sentences used in similar situations, and by 'similar pattern', I meant sentences with different proper nouns. (otherwise same sentences)

"I was born in Canada" - "I was born in France" "Tom loves Mary" - "Sami loves Layla"

I see you leaving many comments on such Sami sentences, but I've never seen you doing that for "I like ___" type sentences. That would be overkill.

The categories I mentioned are just examples. Their names and numbers could be different. What I'm trying to say is that translating shouldn't be the only way of linking. We should be able to use Tatoeba for monolingual purposes, too.

Having different types of comments might be useful, if only comment information was labeled on sentences themselves. Currently, one has to go to each sentence's page to discover if it has any comments or not. It's not practical. I don't think many people notice comments after they fall to back pages of the comment feed.

jiru commented 5 years ago

@soliloquist-tatoeba I totally agree that Tatoeba should be usable for monolingual purposes too.

Please, let’s not focus on the solution (comments) but the problem. I understand "pattern" but "related" is still unclear to me. What are related sentences? Can you give more examples? What are you trying achieve with related sentences? In what ways that would be useful to you?

soliloquist-tatoeba commented 5 years ago

Related sentences are roughly equivalent sentences that are used in similar situations.

How are you? How are you doing? How is it going? What's up? How's things?

I'm going to sleep. I'm going to bed. I'm turning in. I'm hitting the hay.

Of course there might be some nuances (in terms of formality, for example) between them, but they could be clarified with tags or comments. I find grouping such sentences useful. It would help grasping the subject and choosing the right one depending on the situation.

sacredceltic commented 4 years ago

I strongly support the proposal by soliloquist. The types of links should be customisable for extension to other fields of application and the users should be able to decide, in their user’s parameters, which types they’re interested in.

ckjpn commented 4 years ago

Some Thoughts on This Suggestion

If something like this were to be implemented, here are some things to consider.

Possible New Different Types of Links

This is a list of start with. There are likely additions that could be made to this list and perhaps some of the list items could be merged into one item.

Same or Roughly the Same Meaning:

Related:

Displaying these kinds of links

It might be confusing to list these kinds of links along with the links to translations, so maybe another method of displaying these would be needed.

Some possibilities

Adding these kinds of links

Some kind of system would need to be developed for adding these kinds of links.

Perhaps this could be done with a "special links" drop down select menu that showed the various types of special links would be one way to do this.

soliloquist-tatoeba commented 4 years ago

A short-lived and failed experimental attempt to handle the issue with current tools and design:

https://tatoeba.org/eng/sentences/show/8779062

Here are some more examples on the dev website:

soliloquist-tatoeba commented 4 years ago

A draft visualization of the idea after some discussion on the dev site.

related links