How raghava-2812/adaptive-full-triple-matcher works
Full-Triple-Matcher is a research project for integrating heterogeneous knowledge graphs. Its documented workflow loads two graphs, calculates predicate functionality, generates entity and predicate candidates, compares triples, updates pair probabilities, thresholds matches, and evaluates results; triple matching can also run separately. The graph includes the sampled semantic-similarity layer and documented alignment/evaluation branches. Some README-described A-FTM modules are absent from the supplied tree, so they are not represented as source nodes; implementation details beyond the excerpts are uncertain.
Main entry points: FTM - early stop.ipynb Match Thresholding · Calculate triple similarity.ipynb Standalone Triple Matching · dataset KGs and Gold Data · graph_similarity.py Value Similarity
Components 16
KG Inputs
- KGs and Gold Data
dataset
Matching
- Load KG Pair
codes/entity-matching/FTM - early stop.ipynb - Predicate Statistics
codes/entity-matching/FTM - Calculate inverse functionality.ipynb - Candidate Generation
codes/entity-matching/FTM - early stop.ipynb - Full-Triple Comparison
codes/entity-matching/FTM - early stop.ipynb - Pair Probability Update
codes/entity-matching/FTM - early stop.ipynb - Match Thresholding
codes/entity-matching/FTM - early stop.ipynb - Entity Alignment
codes/entity-matching - Schema Alignment
codes/entity-matching - Standalone Triple Matching
codes/triple-matching/Calculate triple similarity.ipynb
Evaluation
- Match Results
codes/entity-matching/Analyze paris and ftm matches.ipynb - Hit@K Results
codes/entity-matching/Analyzing top@n.ipynb - Triple Similarities
codes/triple-matching/Calculate triple similarity.ipynb
Similarity Layer
- Value Similarity
codes/entity-matching/graph_similarity.py - Semantic Type Evidence
codes/entity-matching/semantic_type_matcher.py
Other components
- Researcher
Connections 16
- Researcher → Load KG Pair: runs workflow
README.md - KGs and Gold Data → Load KG Pair: supplies graphs
README.md - Load KG Pair → Predicate Statistics: precedes statistics
README.md - Predicate Statistics → Candidate Generation: precedes candidates
README.md - Candidate Generation → Full-Triple Comparison: precedes comparison
README.md - Full-Triple Comparison → Pair Probability Update: updates probabilities
README.md - Pair Probability Update → Match Thresholding: precedes threshold
README.md - Match Thresholding → Match Results: evaluates matches
README.md - Match Thresholding → Hit@K Results: evaluates ranking
README.md - Researcher → Standalone Triple Matching: runs optionally
README.md - KGs and Gold Data → Standalone Triple Matching: provides triples
README.md - Standalone Triple Matching → Triple Similarities: produces similarities
README.md - Value Similarity → Semantic Type Evidence: uses type evidence
codes/entity-matching/graph_similarity.py - Value Similarity → Full-Triple Comparison: scores values
- Match Thresholding → Entity Alignment: yields entity pairs
README.md - Researcher → Schema Alignment: uses alignment tools
README.md