Benchmarks for software clone detection: A ten-year retrospective
Bibliographic record
Abstract
There have been a great many methods and tools proposed for software clone detection. While some work has been done on assessing and comparing performance of these tools, very little empirical evaluation has been done. In particular, accuracy measures such as precision and recall have only been roughly estimated, due both to problems in creating a validated clone benchmark against which tools can be compared, and to the manual effort required to hand check large numbers of candidate clones. In order to cope with this issue, over the last 10 years we have been working towards building cloning benchmarks for objectively evaluating clone detection tools. Beginning with our WCRE 2008 paper, where we conducted a modestly large empirical study with the NiCad clone detection tool, over the past ten years we have extended and grown our work to include several languages, much larger datasets, and model clones in languages such as Simulink. From a modest set of 15 C and Java systems comprising a total of 7 million lines in 2008, our work has progressed to a benchmark called BigCloneBench with eight million manually validated clone pairs in a large inter-project source dataset of more than 25,000 projects and 365 million lines of code. In this paper, we present a history and overview of software clone detection benchmarks, and review the steps of ourselves and others to come to this stage. We outline a future for clone detection benchmarks and hope to encourage researchers to both use existing benchmarks and to contribute to building the benchmarks of the future.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".