Hacker News Reader

Comment by wkrause13 | Hacker News Reader

The short answer is no, because I don't know enough about those languages to deliver a valuable experience.

I think the value from Langliter comes from two areas:

1. The NLP processing and tooling built on top of that processing 2. Its superior mobile experience

If I just used a simple form of tokenization, I think I could very quickly deliver a superior mobile experience than what is currently available from similar apps. However, I'm not sure if PoS tagging and some of the other techniques I'm using translate to those languages or if they do, how much value they'd add to the learning process.

But it's something I'll certainly revisit if the app ends up being popular and I have the resources to do it properly.

View on HN

Awesome work! I am learning a language and have been fantasising about a tool like this.

What about support of non-latin alphabet languages? I'm learning Dari at the moment. It uses Arabic alphabets, the words are a string of alphabets, there is tenses, verbs, nouns, adjectives. It's different to English only by the alphabet used, the vocabulary and the ordering of the verbs/nouns/subject.

Is it possible for it to be opened up to third parties to contribute new languages?

It's definitely something I hope to get to explore. Langliter is only a side project at the moment, so I'm focusing on languages that I have a little experience with. But as long as the licensing is compatible, I've built Langliter to be pretty flexible in terms of the types of tags it could support.

At least, as far as Korean is concerned, there is a version of the Penn Treebank for Korean: https://repository.upenn.edu/cgi/viewcontent.cgi?article=102...

And of course there are the Universal POS tags, although I'm not sure if they are precise enough for your intended purpose.

Well if you ever get a chance to do this for Indonesian (uses the roman alphabet), I'll deff use the app or at least be willing to try.

As someone living in Indonesia for about two years who has tried other apps, buying grade school books, nothing is interesting to me once the novelty wears off, and my incentives for learning more than navigational stuff and greetings is pretty much nil.

This is a great tool! Can I ask what kind of database and backend setup you're using?

Do you choose articles, then run nlp over them? How do you get dictionary results?

I really like being able to quickly and easily tag words for learning. Thanks. I'm a polyglot and completely relate to the premise of your app and system.

I used the iOS version but an curious, are these native? React Native?

I'm running it on Google Container Engine, Go for the main API layer and CockroachDB for the DB. I actually have a blog post in the works for why I chose CockroachDB. The tl;dr Langliter is offline first so most of the data is synced in the background. Therefore some additional latency makes no real difference and the resilience of a CockroachDB cluster is really impressive.

And good eye, it is React Native app. I should probably look into following platform style guides eventually ( I wouldn't count UI design as a real strength on mine :-) )

Reply on news.ycombinator.com