0.25
0.5
0.75
1.25
1.5
1.75
2
Semi-supervised Instance Matching Using Boosted Classifiers
Published on Jul 15, 20151563 Views
Instance matching concerns identifying pairs of instances that refer to the same underlying entity. Current state-of-the-art instance matchers use machine learning methods. Supervised learning syste
Related categories
Chapter list
Semi-supervised Instance Matching Using Boosted Classifiers00:00
Instance Matching - 100:20
Instance Matching - 201:03
Primary Application: Linked Open Data01:24
Two key aspects - 101:57
Two key aspects - 203:18
An Instance Matching pipeline04:03
Machine learning perspective04:40
But...manual labeling is expensive05:13
How can we minimize labeling effort... - 105:46
How can we minimize labeling effort... - 206:17
How can we minimize labeling effort... - 306:42
Semi-supervised learning07:24
Boosting (specifically Adaboost)09:04
Adaboost algorithm10:35
Proposed System11:10
Proposed Algorithm12:24
(Very high-level) Intuition13:40
Evaluation: Test Cases15:10
Evaluation: Metrics16:21
Evaluation: Baselines16:41
Results: Random Forest17:52
Results: Multilayer Perceptron19:01
Results: External Baselines19:55
Key Takeaways20:37
Conclusion21:28