New platform finds hidden bugs in graph databases
GraphWeave uses an integrated testing approach to uncover previously unknown defects in the NeuG database.
Researchers built GraphWeave, an integrated testing platform for graph databases. It uncovered 19 previously unknown bugs in the NeuG database. All 19 bugs were confirmed.
The team integrated test capabilities from 8 testing methods. They conducted an empirical study to find two blind spots. They found that 12 hours of testing produce thousands of bug reports. Most reports are false positives or duplicates.
The study also found that many equivalence-based metamorphic test cases lose the ability to detect defects. This happens because of the optimization mechanism. GraphWeave uses a multi-agent system to analyze the flood of bug reports. It produces gating rules for issue filtering and adapter rules for dialect adaptation.
GraphWeave also uses plan-aware triage. This routes metamorphic pairs by plan. It skips metamorphic pairs that share a plan on the same graph state.
A graph database is foundational infrastructure for connected data. A metamorphic test case is an equivalence-based test. An optimization mechanism is a process that affects how tests detect defects.
Key facts
- GraphWeave uncovered 19 confirmed bugs in the NeuG database.
- Twelve hours of testing can produce thousands of bug reports.
- Most bug reports are false positives caused by dialect or pattern-matching differences.