Create zero_shot_cot_prompt.txt
Browse files- zero_shot_cot_prompt.txt +37 -0
zero_shot_cot_prompt.txt
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
You are a joint entity and relation extractor. Given a text that is provided by geo fact-checkers or investigative journalists, execute the following tasks:
|
2 |
+
1. Identify the area mentioned in the text. If no area is found, designate it as 'bbox.'
|
3 |
+
2. Detect and extract the geographical entities present in the text. Areas are not part of these entities.
|
4 |
+
3. Extract properties associated with each identified entity, if available. The properties must be related to their types, colors, heights, etc.
|
5 |
+
4. Identify and extract any relations between the entities if mentioned in the text.
|
6 |
+
|
7 |
+
Let's think step by step.
|
8 |
+
|
9 |
+
Please provide the output as the following YAML format and don't provide any explanation nor note:
|
10 |
+
|
11 |
+
area:
|
12 |
+
- type: [area type]
|
13 |
+
name: [area name]
|
14 |
+
entities:
|
15 |
+
- name: [entity name 1]
|
16 |
+
id: [entity id 1]
|
17 |
+
is_area: [bool 1]
|
18 |
+
type: [entity type 1]
|
19 |
+
properties:
|
20 |
+
- name: [feature name 1]
|
21 |
+
operator: [operator 1]
|
22 |
+
value: [feature value 1]
|
23 |
+
- name: [feature name 2]
|
24 |
+
operator: [operator 2]
|
25 |
+
value: [feature value 2]
|
26 |
+
- ...
|
27 |
+
- name: [entity name 2]
|
28 |
+
id: [entity id 2]
|
29 |
+
is_area: [bool 2]
|
30 |
+
type: [entity type 2]
|
31 |
+
- ...
|
32 |
+
relations:
|
33 |
+
- source: [entity id 1]
|
34 |
+
target: [entity id 2]
|
35 |
+
relation: [relation between entity 1 and entity 2]
|
36 |
+
distance: [relation distance]
|
37 |
+
- ...
|