Datasets:

Modalities:
Text
Formats:
csv
Languages:
English
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
Seacow commited on
Commit
b1ae693
1 Parent(s): 6c9caa4

update readme

Browse files
Files changed (1) hide show
  1. README.md +51 -1
README.md CHANGED
@@ -9,4 +9,54 @@ This repository contains the dataset to the paper ["Causal Reasoning of Entities
9
 
10
  # Files
11
  - `data_dev_v2.json` is the development set of CREPE.
12
- - `data_test_v2.json` is the test set of CREPE.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
 
10
  # Files
11
  - `data_dev_v2.json` is the development set of CREPE.
12
+ - `data_test_v2.json` is the test set of CREPE.
13
+
14
+ ---
15
+
16
+ # Explanations of the CREPE dataset
17
+ There are 6 columns in the dataset, namely `goal`, `steps`, `event`, `event_answer`, `entity`, `entity_answer`.
18
+
19
+ - `goal` denotes the goal of a procedure.
20
+
21
+ - `steps` is a list containing all steps involved in a procedure.
22
+
23
+ - `event` is an event whose likelihood of happening change due to the events in steps.
24
+
25
+ - `event_answer` is the ground truth likelihood change. See below for glossary.
26
+
27
+ - `entity` is the entity that directly relates to the `event`. Its state change will have a direct impact on the likelihood of the `event`
28
+
29
+ - `entity_answer` is the ground truth entity state change. See below for glossary.
30
+
31
+ # Glossary
32
+ 0 -> `event`/`entity state` is _less likely_ to happen comparing to the previous step.
33
+
34
+ 1 -> `event`/`entity state` is _equally likely_ to happen comparing to the previous step.
35
+
36
+ 2 -> `event`/`entity state` is _more likely_ to happen comparing to the previous step.
37
+
38
+ ---
39
+
40
+ # Demonstration
41
+
42
+ `goal`: Clean up kitchen counter
43
+
44
+ `steps`:
45
+ 1. Wear rubber gloves.
46
+ 2. Get towels and wipes.
47
+ 3. Use wipes to wipe kitchen counter.
48
+ 4. Use towels to clean kitchen counter.
49
+ 5. Store the gloves.
50
+
51
+ `event`: The likelihood that "_My skin makes contact with things I touch_" after the execution of each step.
52
+
53
+ `ground truth answers`:
54
+ 1 --> "equally likely"
55
+ 1 --> "equally likely"
56
+ 0 --> "less likely"
57
+ 1 --> "equally likely"
58
+ 1 --> "equally likely"
59
+ 1 --> "equally likely"
60
+ 1 --> "equally likely"
61
+ 1 --> "equally likely"
62
+ 2 --> "more likely"