掲示板 Forums - Crowdsourcing Project: Sentence/Grammar Markup

**This is not yet active - still in the planning stages**

After a very positive discussion on renshuu's discord, I've started to prepare for a way to mark (or rather, verify) grammar elements in sentences. That would be these:

These are auto-generated by renshuu, and while they are pretty good, they are not perfect, and it is often the case that multiple expressions might be suggested. (Just think of all the usages of the で particle).

Having a crowdsourcing tools where a user could see a sentence, a few highlighted words, and then "which of the grammar expressions below matches this?" (choose one, or mark I don't know). would be great. We could not only get better sentence data, but we could expand out the examples listed on grammar pages, and even consider expanding out the question database for grammar questions!

I already have the layout in my head - it would look similar to the sentence/word markup tool already available:

Except it would highlight the grammar, and have a list of one or more meanings for that grammar.

Functionally, it's easy to implement. However, There's one, big problem.

Even for the rarer grammar expressions, there are possibly 100s of sentences with the grammar potentially. With something like で, it's well over 10,000. Even restricting it by sentences with audio, native verified, etc, it's still a lot.

So the whole "randomize it" will make it so that no sentence will ever get enough eyes on it to get validated. And this is just for a single grammar expression (out of a 1,000).

My inclination is to do this, instead (but I'd love feedback) - limit the number of validated examples per grammar to something like 20 sentences (through this system). So say that で (period of time) gets 20 validated sentences, then it would not appear in the system anymore.

If a category (say, N5) got 20 for all expressions in the system, then we could move onto the next rank, 40 sentences.

Additionally, instead of randomizing the order of the sentences presented, give *everyone* the same order of sentences for any given grammar, making progress and efficiency much higher.

Honestly, if we could get 20 new examples * 1,000 expressions (so, 20,000 new sentences for the examples and possibly the quiz system), that would be amazing.