MR-based behavioral adequacy assessment for UI component test suites

MR-based behavioral adequacy assessment for UI component test suites This archive contains the code, data, and pre-computed outputs used in an empirical study on behavioral adequacy in UI component test suites. The study models component behavior using metamorphic relations (MRs) and measures how well existing tests exercise and explicitly validate those behaviors. For each component, the corpus includes three artifacts: implementation code (source.js), unit tests (test.js), and an API/documentation excerpt (description.txt). Given these inputs, the pipeline infers component-specific MRs under a fixed six-category UI behavior taxonomy and aligns existing tests with the inferred relations using a hybrid procedure that combines LLM-based semantic analysis with local lexical and assertion checks. The artifact distinguishes Touch, where a test exercises the behavior described by an MR, from Cover, where the test also contains an oracle that validates the expected behavioral relation. Dataset The dataset contains 214 components from four UI component libraries: Material UI, Ant Design, Element Plus, and MUI Base UI. Components span five common UI categories: Inputs, Navigation, Feedback, Data display, and Layout. Component-level samples are stored under: data/test/{id}-{library}-{Category}-{ComponentName}/ Taxonomy. The MR taxonomy used in the study has six top-level categories: input_prop_relations state_event_relations interaction_accessibility_relations visual_layout_relations composition_context_relations data_flow_relations Pipeline The main implementation is under backend/mr_framework/ and includes MR inference, test--MR alignment, completeness measurement, issue-theme alignment, and test-improvement suggestions. Setup cd backend pip install -r requirements.txt cp ../.env.example ../.env # configure DeepSeek / OpenAI / Gemini API keys python3 scripts/run_batch.py python3 scripts/run_batch.py --sample-id=1-mui-material-Inputs-Autocomplete python3 scripts/run_batch.py --ablation-model gpt-5-mini --separate-by-model Keywords Metamorphic relations, UI component testing, Behavioral adequacy, Test adequacy, Test oracles, LLM, Software testing, React/Vue component libraries