Metadata Update June 28, 2024

[Fuga]

Hey Everyone!

Today we would like to announce that we have updated the Search metadata (terms used to return results in search) to make the Search feature far more useful!

---

Searching for Grammar points

Up until this update, searching for a grammar point on Bunpro may have not been the best experience, since you had to know exactly what you were searching for, such as まえに, あとで, なきゃ, etc. There may have been times where you weren’t able to find the grammar point you were looking for, or the style of grammar that you were wanting to learn.

However, as of now, you can search for grammar points simply by typing in relevant terms. Let’s take a look below for some examples of what we mean by relevant terms.

---

New metadata

Relevant terms are what we would usually classify things with by category, such as conjecture, politeness, hearsay, desire, counters, and even parts of speech such as particles and auxiliary verbs.

Here are some examples of what this means!

BEFORE

Before, if you wanted to find the grammar point みたい, you had to search みたい or mitai to pull up the grammar point みたい.

AFTER

Now, as long as you know the usage, the part of speech, translation, or the politeness of a grammar point, you can not only find the grammar point you are looking for, but also other grammar points that share some common meaning with the grammar point you are searching for!

Typing ‘Counter’ will bring up all grammar points that are used in relation to numbers.

Typing ‘Emphasis’ will bring up all emphatic grammar points.

Typing ‘Particle’ will bring up all particles (you can even go further than this and search by type of particle as well!).

Typing common translations of the grammar points in English will also bring up results!

This is just a few small examples, but there are many things that will now bring up great results! We will also be continuing to add more search terms in the future and fill out any ones that we may have missed in the first place.

---

Thank you!

We hope that you find this update useful! We believe that it will make searching for grammar points and finding similar grammar points much easier!

We would appreciate any and all feedback about this feature so that we can keep improving it over time. As always thank you for your continuous support! This update is part of a larger ongoing overhaul we are doing to ‘Search’ on Bunpro, and you can expect even more quality of life improvements in the near future!

I hope that every one has an amazing day!

Fuga

[AshuRage]

Omg that’s a huge one, thank you! Does it also support fuzzy search?

[wct]

Very exciting feature upgrade!

I think it would be greatly helpful if all of the new search features were documented somewhere. It doesn’t have to be fancy, but it would be really great if it was complete, meaning that each existing feature is included in the documentation, even if the ‘documentation’ for a feature is currently minimal.

A good example of such kind of simple search-feature documentation is the “advanced search options documentation” at the Jisho site. E.g. it includes all of the various available special tags that can be used in the search textbox, like #on-mim (for onomatopeia), and explains how to use ‘wildcards’ in the search, like the * and ? special characters.

Or, perhaps in a future update, consider developing an ‘advanced search’ page which contains all the search features as visible user-interface elements. A good example of this would be the “Advanced Search” page on Google Search.

[InsaneSlightly]

Great update as always! Love what you’re doing.

[SeimanDouman]

To be honest, I mainly use the search feature to figure out what a grammar point I find in the wild means.

[Asher]

@wct

A good example of such kind of simple search-feature documentation is the “advanced search options documentation” at the Jisho site. E.g. it includes all of the various available special tags that can be used in the search textbox, like #on-mim (for onomatopeia), and explains how to use ‘wildcards’ in the search, like the * and ? special characters.

Something similar to this is one of the other parts of the full Search overhaul, so is only a matter of time away .

[jrmr50]

Awesome! This will help me a lot. I often use the grammar search.

[Atheon]

Great work, thank you very much!

[Markapi_Suh]

That will be a really useful feature! Thanks for the update

[SimVer]

Thank you!

[JandroSantiago]

This is awesome!

@Fuga

• Typing ‘Counter’ will bring up all grammar points that are used in relation to numbers.
• Typing ‘Emphasis’ will bring up all emphatic grammar points.
• Typing ‘Particle’ will bring up all particles (you can even go further than this and search by type of particle as well!).
• Typing common translations of the grammar points in English will also bring up results!

This is just a few small examples, but there are many things that will now bring up great results!

It would be nice to have a few handy buttons display when you pull up the search page to allow one to browse the grammar points by these categories, especially if they don’t even know what cool terms they could search by!

[eclipse77x]

This is so awesome! Do vocabulary terms come up in the search now too?

[Fuga]

Vocabulary does not show up since grammar and vocabulary is separate!

[entropyofchaos]

Hi, just wanted to give some feedback that search on the web interface is acting odd for me. I was just trying to search for よう as in ように and other grammar that uses this. The search comes back empty. But when I did a search just using the browser’s built in search, I could find it in the list. The search seems to return back fine on the app. Not sure if the update did something that made search not work as expected on the website.

[Fuga]

Hey @entropyofchaos ! This issue has been fixed!

[killhour]

This is huge and relates something I’ve been thinking about a lot lately thanks to the other thread on SRS timing algorithms. Having well-maintained metadata and relationships between grammar points enables much more than search. Depending on how this is implemented, it may be possible to correlate strengths and weaknesses based on correct/incorrect answers. For instance, if a user is working through the SRS sentences for たがる (JLPT N4) | Bunpro and gets them correct until they get to 「彼かれはパソコンを買かいたがって、週末しゅうまつも働はたらいた。」, where they consistently put in answers like 「買かってがる」, that indicates that they potentially aren’t having a problem with たがる directly, but with understanding how that structure gets conjugated to て form. A human teacher would be able to recognize this and give more exercises around using て form with different auxiliary verbs, especially if the student has shown to have this problem with other auxiliary verbs in the past.

Having each example sentence tagged with all the auxiliary verbs, conjugations, etc. would let the system find more examples that show the same situation, even if they aren’t on the same grammar point. Then if you are struggling with a specific combination, the system could suggest other similarly structured sentences to help get context from a broader set of examples instead of just trying to memorize the specific combination for that sentence.

Full disclosure and/or context: my day job is in data model design, especially around search and semantic relationships (taxonomies, ontologies, etc.). I love Bunpro and am more than happy to talk in depth about design approaches and implementation details if it will help improve the platform.