Awesome, and yes, totally makes sense -- you are more learner-centric that way.
Having the full sentence context is actually one of the things I have been thinking about a lot -- this helps both the learner as well as the POS detection in Stanza. I always decided against, because I wanted to build agnostic flash-cards.
However, as your approach allows on-the-fly generation of flash cards, you always stay close to the learner progress. I could (e.g.) pick some Gutenberg fairy tales, allow the learner to read them in their target language and provide bi- and omni-directional translations across all languages. Creating flash cards from the source material keeps the learner in progress (context), allows to learn new words step-by-step (discovery), as well as providing a fun learning experience and measurable progress. Similarly, instead of fairy tales, we could use some series in combination with its subtitles. This allows video-progress. Awesome x2!
Sidenote: The awesome part about HN is that I get to chat with like-minded people and directly grasp some new inspiration. Probably I ought to visit some in-person hacker spaces :)
- How do you transform and enrich the data? How does your pipeline look?
- What are your key challenges?
- Which tools do you use? What is your 'stack'? (Stanze, wordfreq, Whisper, wn, ...)
Background: I am currently building a multi-lang vocabulary hub for language learning. The goal is to match core words/lemmas to their senses/concepts, and then be able to generate multi-language flash cards.
I am still stuck on the sense alignment and fingerprinting (example: should 'to shop', 'einkaufen', ' alışveriş yapmak' and 'go shopping' point to the same concept of 'shop'?), but in a later stage I want to allow user-submission and data enrichment for IPA, pictograms [1] and audio.
Use-case (the dream): I come back from language class, I input new vocab and I output new Anki cards that work across all my fluent languages.
Currently, I mostly find myself knee-deep in problems of linguistics, NLP, Python and getting an LLM to do exactly what I want. At the same time it is a super fun project, and really makes me feel the joy of programming again. LLMs are magic, time just flies by, and all the random projects I always wanted to do suddenly materialize.
For coding, I mostly use free Gemini and some deepseek-v4-flash via openrouter to keep a tight oversight and understand the problem space. Maybe this slows me down, but agentic code jsut does not align with me. Overall, I haven't spent more than 2 € in total.
So far, surprisingly, the biggest problem is the lack of high-quality, free input data (example: English has the Oxford 5000 words as core vocabulary, but it is difficult to find the same for e.g. Turkish).
2nd place is the lack of high-quality synsets/wordnets (cross-language is mostly incomplete), and the 3rd place is getting LLMs to reliable play to their strength (on paper, a LLM is the perfect tool to provide multi-lang sense equivalents)
I plan to do a full writeup sometimes, but first I need it to work :)
There was another article/blog on hackernews some time ago along the lines of 'I'm an editor, here is how I edit my friends texts' with some really good advice.
Unfortunately I can't find it anymore -- if someone knows which post I mean, I'd appreciate sharing it with me again.
Interesting approach. I am currently looking for jobs and went the 'career coaching' route for my CV. I did a few iterations with my coach until I got my current result (ideally I had a link):
I first looked at Canva templates, but apparently nowadays you are supposed to do black/white and no fancy designs for ATS readability. Then I tried it with Google Docs b/w resumee template, which kinda got me to write actual skills. Then I approached the coach, got her template and iterated, and then I also added some rules from here (https://principiae.be/pdfs/ECV-1.01.pdf).
I also involved ChatGPT to analyze job postings and to get the mix of keywords in my resumme right. Tools like https://tagcrowd.com/ also help with that. For example, I am targeting 'IT analyst' roles, and it does make sense that I have the word 'analysis' a few times in my CV.
E: mine is basically structured the following way
Name
Title
Summary
3x5 ATS keywords/skills specific to my profile and role
last ten years, also written in a way that 'gamifies' ATS: 'Year, worked as ROLE at Company, did XYZ'
--page 2--
Education (degree + grades)
Skills Training
Languages
Some more IT skills (programming languages, project management, ...)
E2: I obviously have no idea what I am doing, but I got three interview proposals for 10 applications, so I guess 30%.
Imo the problem with Org-mode is that it "afterloads" all the complexity into the :PROPERTIES: drawers. Those quickly transform into unreadable hideous beasts of metadata that _require_ you to use Emacs. Every plugin developer likes to create their own set of properties, which directly leads to the problem of "multiple Markdown flavours". Extensibility is not always good.
Interesting that this has a name. I've been doing this naturally for quite some time now without knowing the science behind it; it just feels "right" for me.
Thinking about it, whenever I had a job where I ate lunch with coworkers (besides my breakfast+dinner combo) I noticed that I gained lots of weight.
I've been intensively using Orgmode for a year and a half (wrote my thesis in it) and then abandoned it to switch to Markdown.
Orgmode quickly turns into not-quite-plaintext with humanly impossible to read and very distracting data structs stuck inside the text. I think these were called "Properties"?
For me the beauty of plain-text is that it can be read and written in any Editor without needing syntax highlighting. Here Orgmode fails for me, as I found it unreadable and unusable without an Emacs-esque toolkit.
I am a NixOS user, but am interested in Guix. Based on a cursory look I have some questions:
- How big is Guix on GNU? Does it throw wrenches in your way if you do anything "unfree"? How easy is it to install the nvidia-drivers?
- How is the package ecosystem? 20.000 official packages seems a little low? Are there community packages? How easy it is to create your own packages? Why is the Neovim package only at version 4.4 [1]? Isn't version 5 officially released?
- How good is the documentation?
- Is there a project roadmap? Are there any issues preventing adoption?
Having the full sentence context is actually one of the things I have been thinking about a lot -- this helps both the learner as well as the POS detection in Stanza. I always decided against, because I wanted to build agnostic flash-cards.
However, as your approach allows on-the-fly generation of flash cards, you always stay close to the learner progress. I could (e.g.) pick some Gutenberg fairy tales, allow the learner to read them in their target language and provide bi- and omni-directional translations across all languages. Creating flash cards from the source material keeps the learner in progress (context), allows to learn new words step-by-step (discovery), as well as providing a fun learning experience and measurable progress. Similarly, instead of fairy tales, we could use some series in combination with its subtitles. This allows video-progress. Awesome x2!
Sidenote: The awesome part about HN is that I get to chat with like-minded people and directly grasp some new inspiration. Probably I ought to visit some in-person hacker spaces :)