You need to agree to share your Hugging Face information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Final Release: Items with Reproducible Knowledge Sources

中文版:README.md

This release contains 5,551 items. items/china, items/us and items/india are each provided as .jsonl and .csv. Fields are described in data_dictionary.csv.

File Selected-response Constructed-response Total
china 1,735 120 1,855
us 1,782 120 1,902
india 1,794 0 1,794

This directory contains only fields that can be derived from the knowledge sources. It does not include the identifiers, years, modules, tracks or knowledge points of the original exams. Item IDs are newly assigned and carry no information about the original items. The items were originally written to cover the knowledge tested by teacher exams. This directory does not reflect that link, so refer to the methods section of the paper when using or citing the data.

Each item records one knowledge source (source_url). The source text is not included, so open the URL to read it. Web pages may change over time.

Questions, options, reference answers and scoring rubrics of Chinese items remain in Chinese. license, legal_basis and legal_text each have an English column (_en). English versions of Chinese legal text are unofficial translations, and the original text prevails.

Source verification

Of the 5,551 items, 1,095 were reviewed by a person. Of those, the answer to 525 can be derived from the question and options alone, and the support for 12 is weak. The sources for the other 4,456 items were selected by a program and have not been confirmed by a person, so the source alone may not be enough to choose the correct answer.

In a manual check of 100 items whose correct option appeared verbatim in a program-selected excerpt, the excerpt alone was not enough to choose the answer for 47.

Source licenses

The licenses below apply to the source pages, not to this dataset.

license_en Items Basis (legal_basis_en)
CC BY-SA 4.0 4,527 Page footer statements on Wikipedia and Wiktionary
Not protected by copyright law 979 Copyright Law of the People's Republic of China (2020 Amendment), Article 5, Item (1)
Public domain in the U.S. 30 17 U.S.C. §105(a), §101
Public domain (copyright term expired) 12 Copyright Law of the People's Republic of China (2020 Amendment), Article 23, Paragraph 1
CC BY-ND 3.0 TW 2 Statement on the public licensing site for the Mandarin dictionaries of the Ministry of Education (Taiwan)
PSF-2.0 1 Official Python "History and License" page
  • Of the 979 items marked "Not protected by copyright law", 463 cite curriculum standards issued by the Chinese Ministry of Education. The notice issuing them is an administrative document, but no court ruling confirms that the full text of each subject standard falls under Article 5.
  • Some Chinese government websites restrict reposting of their content. This dataset only links to those pages. If you copy page content, follow the site's statement and the applicable license.
  • Items whose only sources were restricted pages (Baidu Baike, training-company websites, textbook sample chapters and similar) are not in this dataset.
Downloads last month
-