Junyi Academy Online Learning Activity Dataset
The Junyi Academy Online Learning Activity Dataset is a large log of mathematics practice from Junyi Academy, a Taiwanese non-profit online learning platform run by the Junyi Academy Foundation. Published on Kaggle in 2020, it records 16,217,311 problem attempts made by more than 72,000 students across 1,330 mathematics exercises over one school year, from August 2018 to July 2019. Alongside the attempt logs it ships an expert-built knowledge map: each exercise carries a topic and area, a position in the map and, importantly, its prerequisite exercises, giving an annotated graph of how one skill depends on another.
You access it by downloading a set of CSV files from Kaggle (a free account is needed). Log_Problem.csv holds the attempt records, while separate content and metadata files describe the exercises and the exercise relationships. Timestamps are rounded to the nearest quarter hour, which limits fine-grained sequence reconstruction but does not affect coarse ordering.
For an AI tutor this is one of the few public datasets that pairs real practice behaviour with an explicit prerequisite structure, so it suits knowledge tracing and mastery modelling (estimating what a learner knows from their attempt history) and, through the prerequisite graph, curriculum sequencing: deciding what to teach next. For RAG the logs themselves are behavioural rather than textual, so the retrievable content is the concept and prerequisite graph, useful for grounding recommendations about learning order.
The licence is Creative Commons Attribution-NonCommercial-ShareAlike 4.0. This is non-commercial: you may not use the data in a commercial product, derivatives must carry the same licence, and attribution is required; Junyi's trademarks are excluded. Anyone building a paid tutor should treat it as research and evaluation only.
It complements the text-first education sources we list. Where OpenStax, CK-12, Siyavula and Khan Academy supply teachable content, and Common Core and NGSS supply standards, Junyi supplies the missing behavioural layer: how real students moved through a structured maths curriculum.
Related sources
Achievement Standards Network
Machine-readable curriculum standards from US states, national bodies and other jurisdictions, modelled as an RDF graph of URI-addressable learning objectives with cross-jurisdiction alignments. Now run by D2L and free to use.
AGIEval
8,062 questions drawn from 20 official standardised exams (SAT, LSAT, GMAT, GRE, Gaokao, AMC/AIME and more) in English and Chinese, packaged as a human-centric benchmark for evaluating foundation models.
CASE Network (1EdTech)
A public registry of machine-readable learning-standard frameworks from all 50 US states and other issuing agencies, run by 1EdTech in the CASE JSON format. The digitally referenceable spine of what to teach, at which level and in what order, rather than the teaching content itself.
CEFR Companion Volume Descriptors
The Council of Europe's 2020 CEFR Companion Volume descriptor set, the de facto international standard for levelling language proficiency from Pre-A1 to C2. A spine of can-do statements for sequencing and grading language teaching.