Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -10,7 +10,7 @@ language:
|
|
10 |
- sk
|
11 |
tags:
|
12 |
- sentiment
|
13 |
-
-
|
14 |
- parliament
|
15 |
- parlament
|
16 |
pretty_name: ParlaSent
|
@@ -27,10 +27,23 @@ size_categories:
|
|
27 |
|
28 |
### Dataset Summary
|
29 |
|
|
|
|
|
30 |
The dataset consists of five training datasets and two test sets. The test sets have a _test.jsonl suffix.
|
31 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
Dataset is described in detail in our [paper](https://arxiv.org/abs/2309.09783).
|
33 |
|
|
|
34 |
### Data Attributes
|
35 |
The attributes in training data are the following:
|
36 |
- sentence - the sentence labeled for sentiment
|
|
|
10 |
- sk
|
11 |
tags:
|
12 |
- sentiment
|
13 |
+
- classification
|
14 |
- parliament
|
15 |
- parlament
|
16 |
pretty_name: ParlaSent
|
|
|
27 |
|
28 |
### Dataset Summary
|
29 |
|
30 |
+
This dataset was created and used for sentiment analysis experiments.
|
31 |
+
|
32 |
The dataset consists of five training datasets and two test sets. The test sets have a _test.jsonl suffix.
|
33 |
|
34 |
+
Each test set consists of 2,600 sentences, annotated by one highly trained annotator. Training datasets were internally split into "train", "dev" and "test" portions" for performing language-specific experiments.
|
35 |
+
|
36 |
+
The 6-level annotation schema, used by annotators, is the following:
|
37 |
+
- Positive for sentences that are entirely or predominantly positive
|
38 |
+
- Negative for sentences that are entirely or predominantly negative
|
39 |
+
- M_Positive for sentences that convey an ambiguous sentiment or a mixture of sentiments, but lean more towards the positive sentiment
|
40 |
+
- M_Negative for sentences that convey an ambiguous sentiment or a mixture of sentiments, but lean more towards the negative sentiment
|
41 |
+
- P_Neutral for sentences that only contain non-sentiment-related statements, but still lean more towards the positive sentiment
|
42 |
+
- N_Neutral for sentences that only contain non-sentiment-related statements, but still lean more towards the negative sentiment
|
43 |
+
|
44 |
Dataset is described in detail in our [paper](https://arxiv.org/abs/2309.09783).
|
45 |
|
46 |
+
|
47 |
### Data Attributes
|
48 |
The attributes in training data are the following:
|
49 |
- sentence - the sentence labeled for sentiment
|