Junyi Academy Math Practicing Log (to Jan. 2015) Partial Samples

To meet the size limitation of PSLC DataShop website, this dataset only contains problem log from small portion (1/120) of users. Please go to https://pslcdatashop.web.cmu.edu/DatasetInfo?datasetId=1198 and download a complete dataset from the extra files of the dataset.

Junyi Academy (http://www.junyiacademy.org/) is a Chinese e-learning website which is established on the basis of the open-source code released by Khan Academy (https://www.khanacademy.org/) in 2012.
In its math curriculum, Junyi Academy provides hundreds of exercises with randomly generated numbers like Khan Academy.

Additional Notes:
There is an exercise hierarchy in the dataset: Area -> Topic (Unit) -> Exercise (Section) -> Problem. In Khan Academy exercise framework, exercise contains one or more problem templates. The numbers in problems are randomly generated, so that users would practice problems with different numbers even though they select the same exercise. That is, the same problem in the dataset is generated by the same template with different numbers.

The practice log in Junyi Academy does not record some detail behaviors when users answer a problem. For example:
1. We do not know the order of each answering attempt and each hint request in the same problem.
2. The tutor only records the accuracy of the first attempt, so we are unable to infer whether users provide correct answers eventually if there are multiple attempts.
3. We only records IP instead of session ID.

Therefore, we prepare our transaction data by making following assumptions:
1. In the same problem, users always make all attempts first, and request all hints next.
2. If there are multiple attempts, the last attempt is always false.
3. The same user with same IP would have the same session ID.