Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
json
Sub-tasks:
sentiment-classification
Languages:
Turkish
Size:
100K - 1M
Tags:
sentiment
License:
BayanDuygu
commited on
added benchmark info
Browse files
README.md
CHANGED
@@ -63,10 +63,6 @@ dataset_info:
|
|
63 |
MüşteriYorumları is a Turkish e-commerce customer reviews dataset of size 103K, scraped from Hepsiburada.com and Trendyol.com. These reviews encompass a wide
|
64 |
array of product categories, including apparel, food items, baby products, and books. Review stars are in range of 1-5 stars.
|
65 |
|
66 |
-
|
67 |
-
This dataset is a part of [SentiTurca](https://huggingface.co/datasets/turkish-nlp-suite/SentiTurca) benchmark, in the benchmark the subset name is **e-commerce**, named according to the GLUE tasks.
|
68 |
-
Model benchmarking information can be found under SentiTurca HF repo and benchmarking scripts can be found under [SentiTurca Github repo](https://github.com/turkish-nlp-suite/SentiTurca).
|
69 |
-
|
70 |
The star distribution is as follows:
|
71 |
|
72 |
|
@@ -79,7 +75,7 @@ The star distribution is as follows:
|
|
79 |
| 5 | 30,314 |
|
80 |
| total | 103,920 |
|
81 |
|
82 |
-
The star distribution is quite skewed towards 4+ reviews.
|
83 |
|
84 |
## Dataset Instances
|
85 |
|
@@ -92,13 +88,29 @@ An instance looks like:
|
|
92 |
}
|
93 |
```
|
94 |
|
95 |
-
|
96 |
-
|
97 |
## Data Split
|
98 |
|
99 |
| name |train|validation|test|
|
100 |
|---------|----:|---:|---:|
|
101 |
|MüşteriYorumları Customer Reviews|73920|15000|15000|
|
102 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
## Citation
|
104 |
Coming soon!!
|
|
|
63 |
MüşteriYorumları is a Turkish e-commerce customer reviews dataset of size 103K, scraped from Hepsiburada.com and Trendyol.com. These reviews encompass a wide
|
64 |
array of product categories, including apparel, food items, baby products, and books. Review stars are in range of 1-5 stars.
|
65 |
|
|
|
|
|
|
|
|
|
66 |
The star distribution is as follows:
|
67 |
|
68 |
|
|
|
75 |
| 5 | 30,314 |
|
76 |
| total | 103,920 |
|
77 |
|
78 |
+
The star distribution is quite skewed towards 4+ reviews. For more information about dataset statistics, please refer to the [research paper]().
|
79 |
|
80 |
## Dataset Instances
|
81 |
|
|
|
88 |
}
|
89 |
```
|
90 |
|
|
|
|
|
91 |
## Data Split
|
92 |
|
93 |
| name |train|validation|test|
|
94 |
|---------|----:|---:|---:|
|
95 |
|MüşteriYorumları Customer Reviews|73920|15000|15000|
|
96 |
|
97 |
+
|
98 |
+
## Benchmarking
|
99 |
+
This dataset is a part of [SentiTurca](https://huggingface.co/datasets/turkish-nlp-suite/SentiTurca) benchmark, in the benchmark the subset name is **e-commerce**, named according to the GLUE tasks.
|
100 |
+
Model benchmarking information can be found under SentiTurca HF repo and benchmarking scripts can be found under [SentiTurca Github repo](https://github.com/turkish-nlp-suite/SentiTurca).
|
101 |
+
|
102 |
+
For this dataset we benchmarked a transformer based model BERTurk and a handful of LLMs. Success of each model is follows:
|
103 |
+
|
104 |
+
| Model | acc./F1 |
|
105 |
+
|---|---|
|
106 |
+
| Gemini 1.0 Pro | 1.0/1.0 |
|
107 |
+
| GPT-4 Turbo | 0.64/0.63 |
|
108 |
+
| Claude 3 Sonnet | 0.57/0.53 |
|
109 |
+
| Llama 3 70B | 0.58/0.55 |
|
110 |
+
| Qwen2-72B | 0.53/0.50 |
|
111 |
+
| BERTurk | 0.66/0.64 |
|
112 |
+
|
113 |
+
For a critique of the results, misclassified instances and more please consult to the [research paper]().
|
114 |
+
|
115 |
## Citation
|
116 |
Coming soon!!
|