Overfitting in Semantics-Based Program Repair: IntroClass and Codeflaws Benchmarks

1.0 GB dataset supplements a 2018 research paper on overfitting in semantics-based automated program repair. The data likely contains results from systematic experiments using the IntroClass and Codeflaws benchmarks to characterize patch correctness issues. Author LE DINH, XUAN BACH published the dataset on figshare under a CC-BY-4.0 license.

Use Cases

• Analyze overfitting patterns in semantics-based program repair based on the described benchmark experiments
• Compare overfitting characteristics between semantics-based and heuristics-based APR techniques based on the study's focus
• Validate new program repair methods against established overfitting benchmarks mentioned in the description
• Study the relationship between test suite adequacy and patch generalization based on the dataset's research context

Strengths

• Dataset size is 1.0 GB, indicating substantial experimental results or code artifacts
• Explicitly supplements a peer-reviewed 2018 paper published in Empirical Software Engineering
• Uses established benchmarks (IntroClass and Codeflaws) suited for assessing repair quality
• Released under the permissive CC-BY-4.0 license for broad reuse

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Description metadata is limited; actual data content and structure require manual inspection

Provenance

Source

Research paper by LE, Dinh Xuan Bach; THUNG, Ferdian; LO, David; and LE GOUES, Claire

Collection Method

Likely generated from systematic experiments on IntroClass and Codeflaws benchmarks

Time Range

Associated with research published in 2018

Freshness

Last updated 2026-05-06 09:32:34; freshness should be verified

Geography

null