Steps in EM
1. Loading in the input tables
2. Down sampling
3. Blocking
4. Combining multiple blocker outputs
5. Debugging the blocker output
6. Sampling and labeling
7. Selecting best learning-based matcher using labeled data
8. Generating feature table
9. Extracting feature vectors
10. Debugging matcher output
11. Predicting the matches using trained matcher
12. Evaluating match predictions