Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
json
Sub-tasks:
sentiment-classification
Languages:
Turkish
Size:
10K - 100K
Tags:
sentiment
License:
BayanDuygu
commited on
Commit
•
137d0ff
1
Parent(s):
ba8479e
added benchmark info
Browse files
README.md
CHANGED
@@ -70,10 +70,6 @@ BüyükSinema is a Turkish movie reviews dataset of size 87K, scraped from Sinef
|
|
70 |
|
71 |
This is a merge of the three different datasets from two resources, hence we scaled the output stars into the range of 1-10 accordingly.
|
72 |
|
73 |
-
This dataset is a part of [TRGLUE](https://huggingface.co/datasets/turkish-nlp-suite/TrGLUE) and [SentiTurca](https://huggingface.co/datasets/turkish-nlp-suite/SentiTurca) benchmarks, in the benchmark the subset name is **TrSST-2**, named according to the GLUE tasks.
|
74 |
-
Also the TrGLUE and SentiTurca tasks are binary classification tasks to follow original GLUE conventions. Here, you can access the original star ratings if you want a challenge.
|
75 |
-
Also benchmarking info is given in TrGLUE repo.
|
76 |
-
|
77 |
The star distribution is as follows:
|
78 |
|
79 |
|
@@ -91,7 +87,7 @@ The star distribution is as follows:
|
|
91 |
| 10 | 8,820 |
|
92 |
| total | 87,328 |
|
93 |
|
94 |
-
The star distribution is quite skewed towards 7+ stars.
|
95 |
|
96 |
## Dataset Instances
|
97 |
|
@@ -104,13 +100,19 @@ An instance looks like:
|
|
104 |
}
|
105 |
```
|
106 |
|
107 |
-
|
108 |
-
|
109 |
## Data Split
|
110 |
|
111 |
| name |train|validation|test|
|
112 |
|---------|----:|---:|---:|
|
113 |
|BüyükSinema Movie Reviews|67328|10000|10000|
|
114 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
## Citation
|
116 |
Coming soon!!
|
|
|
70 |
|
71 |
This is a merge of the three different datasets from two resources, hence we scaled the output stars into the range of 1-10 accordingly.
|
72 |
|
|
|
|
|
|
|
|
|
73 |
The star distribution is as follows:
|
74 |
|
75 |
|
|
|
87 |
| 10 | 8,820 |
|
88 |
| total | 87,328 |
|
89 |
|
90 |
+
The star distribution is quite skewed towards 7+ stars. For more information about dataset statistics, please refer to the [research paper]().
|
91 |
|
92 |
## Dataset Instances
|
93 |
|
|
|
100 |
}
|
101 |
```
|
102 |
|
|
|
|
|
103 |
## Data Split
|
104 |
|
105 |
| name |train|validation|test|
|
106 |
|---------|----:|---:|---:|
|
107 |
|BüyükSinema Movie Reviews|67328|10000|10000|
|
108 |
|
109 |
+
## Benchmarking
|
110 |
+
This dataset is a part of [TRGLUE](https://huggingface.co/datasets/turkish-nlp-suite/TrGLUE) and [SentiTurca](https://huggingface.co/datasets/turkish-nlp-suite/SentiTurca) benchmarks, in the benchmark the subset name is **TrSST-2**, named according to the GLUE tasks.
|
111 |
+
Also the TrGLUE and SentiTurca tasks are binary classification tasks to follow original GLUE conventions.
|
112 |
+
In this repo, you can access the original star ratings if you want a challenge.
|
113 |
+
|
114 |
+
We benchmarked the transformer based model BERTurk on the binary classification task, this model achieved a **0.67** Matthews's correlation coefficient.
|
115 |
+
More information can be found in the [research paper]() and benchmarking code can be found under [TrGLUE Github repo](https://github.com/turkish-nlp-suite/TrGLUE).
|
116 |
+
|
117 |
## Citation
|
118 |
Coming soon!!
|