AutoStub Training Dataset

This dataset contains input-output pairs used for training the AutoStub system, which automatically generates symbolic stubs for external functions during symbolic execution.

Dataset Description

The dataset consists of input-output pairs from Java standard libraries, focusing on primitive types and string operations. It was generated using a stratified sampling approach to ensure diverse inputs across different data types.

Dataset Structure

The dataset is organized as a single CSV file with the following columns:

• method_name: The name of the Java method
• sample_id: Unique identifier for each sample (method_name + sample index)
• input: Input values provided to the method (as string representation)
• input_types: Types of input parameters
• output: Output value returned by the method
• output_type: Return type of the method

Method Coverage

The dataset covers methods from the following Java classes:

• java.lang.Boolean
• java.lang.Double
• java.lang.Float
• java.lang.String
• java.lang.Byte
• java.lang.Short
• java.lang.Character
• java.lang.Integer
• java.lang.Long
• java.lang.Math
• java.lang.StrictMath

Dataset Creation

The dataset was created by:

1. Identifying suitable methods in Java standard libraries
2. Using reflection to dynamically invoke methods with random inputs
3. Collecting input-output pairs for each method
4. Applying stratified sampling to ensure diverse inputs

Usage

Loading the Dataset

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Felix6326727/AutoStub")

# Loop through the dataset and print the first 5 samples
for i, sample in enumerate(dataset["train"]):
    if i >= 5:
        break

    print(f"Sample {i}:")
    print(f"Method Name: {sample['method_name']}")
    print(f"Input: {sample['input']}")
    print(f"Input Types: {sample['input_types']}")
    print(f"Output: {sample['output']}")
    print(f"Output Type: {sample['output_type']}")
    print()

Applications

This dataset can be used to:

• Train machine learning models to approximate Java method behavior
• Evaluate the accuracy of symbolic stubs for symbolic execution
• Study the behavior of Java standard library methods across diverse inputs

Citation

If you use this dataset, please cite:

@inproceedings{maechtle2025autostub,
  title={AutoStub: Genetic Programming-Based Stub Creation for Symbolic Execution},
  author={Felix M{"a}chtle, Nils Loose, Jan-Niclas Serr, Jonas Sander, Thomas Eisenbarth},
  booktitle={Proceedings of the 18th ACM/IEEE International Workshop on Search-Based and Fuzz Testing, SBFT 2025},
  year={2025}
}

Source

This dataset is derived from the AutoStub project.