by: Alexander L. Hayes, Kaushik Roy

<< "Getting Started" | BoostSRL Wiki | "Basic Usage Parameters" >>

1. Basic File Structure
2. Advanced File Structure

Files that BoostSRL operates on are stored in a folder with three things:

1. background.txt : Modes
2. train/ folder :

• train_bk.txt : Pointer to the background file.
• train_facts.txt : Facts
• train_pos.txt : Positive examples
• train_neg.txt : Negative examples

3. test/ folder :

• test_bk.txt : Pointer to the background file.
• test_facts.txt : Facts
• test_pos.txt : Positive examples
• test_neg.txt : Negative examples

Example:

File structure for the Cora dataset, notice that the background is called "cora_bk.txt" in this example.

This is okay if train_bk.txt and test_bk.txt both point correctly with: import: "../cora_bk.txt".

Table of Contents | BoostSRL Wiki

After training/testing, more files and folders will appear. This advanced guide explains what each of them are, including the contents of the models, dotFiles, bRDNs, and WILLtheories directories.

Not all of these will necessarily appear, for example: the CombinedTrees(target).dot only appears when the -combine flag is set.

1 :Data/
2 :├── background.txt
3 :├── BoostSRL.jar
4 :├── test
5 :│   ├── query_(target).db
6 :│   ├── results_(target).db
7 :│   ├── test_bk.txt
8 :│   ├── test_facts.txt
9 :│   ├── test_infer_dribble.txt
10:│   ├── test_neg.txt
11:│   └── test_pos.txt
12:└── train
13:    ├── models
14:    │   ├── bRDNs
15:    │   │   ├── dotFiles
16:    │   │   │   ├── CombinedTrees(target).dot
17:    │   │   │   ├── rdn.dot
18:    │   │   │   ├── WILLTreeFor_(target)0.dot
19:    │   │   │   ├── ...
20:    │   │   │   └── WILLTreeFor_(target)9.dot
21:    │   │   ├── (target).model
22:    │   │   ├── (target)_testsetStats_pos_neg_Lits1Trees10Skew2.txt
23:    │   │   ├── old_(target).model
24:    │   │   ├── predictions_pos_neg_Lits1Trees10Skew2.csv
25:    │   │   └── Trees
26:    │   │       ├── CombinedTreesTreeFile(target).tree
27:    │   │       ├── (target)Tree0.tree
28:    │   │       ├── ...
29:    │   │       └── (target)Tree9.tree
30:    │   └── WILLtheories
31:    │       ├── (target)_learnedWILLregressionTrees.txt
32:    │       └── old_(target)_learnedWILLregressionTrees.txt
33:    ├── schema.db
34:    ├── train_bk.txt
35:    ├── train_facts.txt
36:    ├── train_gleaner.txt
37:    ├── train_learn_dribble.txt
38:    ├── train_neg.txt
39:    └── train_pos.txt

1. Data/: Directory that contains the data we train/test on.
2. background.txt: modes file used to guide the search space.
3. BoostSRL.jar: If you're using a jar file, you'll usually keep it at the root of the data directory.
4. test/: Directory containing testing data.
5. query_(target).db:
6. results_(target).db: Results of running inference (testing) on the data.
7. test_bk.txt: Pointer to the background.txt
8. test_facts.txt: Predicates described in the background.txt
9. test_infer_dribble.txt:
10. test_neg.txt: Negative testing examples.
11. test_pos.txt: Positive testing examples.
12. train/: Directory containing training data.
13. models/:
14. bRDNs/:
15. dotFiles/:
16. CombinedTrees(target).dot: Combined Tree of the target, results from using the -combine flag.
17. rdn.dot:
18. WILLTreeFor_(target)0.dot: First of the boosted trees.
19. ...: For each tree, there will be an associated file named WILLTreeFor_(target)#.dot
20. WILLTreeFor_(target)9.dot: Last of the boosted trees, equal to one less than the number of trees learned.
21. (target).model:
22. (target)_testsetStats_pos_neg_Lits1Trees10Skew2.txt:
23. old_(target).model:
24. predictions_pos_neg_Lits1Trees10Skew2.csv:
25. Trees/:
26. Combined TreesTreeFile(target).tree
27. (target)Tree0.tree:
28. ...: For each tree, there will be an associated file named (target)Tree#.tree
29. (target)Tree9.tree:
30. WILLtheories/:
31. (target)_learnedWILLregressionTrees.txt:
32. old_(target)_learnedWILLregressionTrees.txt:
33. schema.db:
34. train_bk.txt: Pointer to the background.txt
35. train_facts.txt: Predicates described in the background.txt
36. train_gleaner.txt:
37. train_learn_dribble.txt:
38. train_neg.txt: Negative training examples.
39. train_pos.txt: Positive training examples.

Table of Contents | BoostSRL Wiki

<< "Getting Started" | BoostSRL Wiki | "Basic Usage Parameters" >>