Skip to content

VOCAB CURATOR

2026.09

Personalized vocabulary curation

MABC 2026 finalistTeam of two · sole developer

Only the words this learner needs, not an average learner's Learner profile Context · level · goals Source text Text · PDF upload SOLAR PRO 4 Select vocabulary Importance · examples Learning history DB Skips words already picked Word cards Meaning · example · importance Use-today phrases 3 per word CSV word list Learner Up to 20 words per text · Node.js · Express.js · PDF.js

Problem

Commercial vocabulary books are written for an 'average learner' who doesn't exist. The words you would actually use get left out, and the example sentences are textbook ones that have nothing to do with daily life.

My part

We passed the preliminaries with my teammate's idea; the final round was about turning it into a working service, and I was the only developer on our two-person team. To turn the preliminary-round skill into a web service, I called Solar Pro 4 directly, passed the source text, persona, and learning history into the prompt, and connected a database that catches duplicate words and tracks the learner profile.

Result

It picks vocabulary from a source text based on the learner's history and profile, then gives each word an importance score, "expressions to use today," and a CSV word list. Team Bulldog (No.45) reached the MABC 2026 finals.
Node.jsExpress.jsSolar Pro 4PDF.jsDocker
GitHub