Back to datasets
Dataset assetOpen Source CommunityCode AnalysisSoftware Defect Detection

LorenzH/juliet_test_suite_c_1_3

This dataset includes all test cases from NIST's Juliet test suite for the C and C++ programming languages. Each sample provides a good and a defective implementation, extracted via the Juliet suite's OMITGOOD and OMITBAD preprocessor macros. The dataset supports software defect prediction and code clone detection tasks. Its structure comprises data instances, fields, and splits. Fields include index, filename, defect class, good code, and bad code. Splits contain training and test set sizes. The dataset is synthetic, with all samples handcrafted, and therefore does not fully represent real‑world software defects.

Source
hugging_face
Created
Nov 28, 2025
Updated
Mar 21, 2023
Signals
362 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Card: Juliet Test Suite 1.3

Dataset Overview

This dataset contains all test cases from NIST's Juliet Test Suite for the C and C++ programming languages. Each sample includes a good and a defective implementation, extracted using the Juliet suite's OMITGOOD and OMITBAD preprocessor macros.

Supported Tasks and Leaderboards

  • Software defect prediction
  • Code clone detection

Languages

C and C++ programming languages

Dataset Structure

Data Instances

Data Fields

IndexNameTypeDescription
0indexintIndex of each sample in the dataset
1filenamestrPath of the test case file, including the filename
2classintDefect category, i.e., the CWE identifier set for the sample
3goodstrSource code of the benign implementation
4badstrSource code of the defective implementation

Data Splits

TypeSize
train80,706 cases
test20,177 cases

Dataset Creation

Source

https://samate.nist.gov/SARD/test-suites/112

Usage Considerations

Societal Impact

Bias Discussion

Other Known Limitations

The Juliet test suite is a synthetic dataset; all samples are handcrafted and therefore do not fully represent real software defects. Applying classifiers trained on these samples to real‑world environments may lead to degraded performance and severe misclassifications, potentially overlooking critical software defects.

Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio