teami12 commited on
Commit
e6c9549
1 Parent(s): 5f059a8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +118 -3
README.md CHANGED
@@ -1,3 +1,118 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - text-classification
5
+ - sentence-similarity
6
+ - translation
7
+ language:
8
+ - en
9
+ - sr
10
+ tags:
11
+ - cross-lingual
12
+ - semantic-text-similarity
13
+ - sts-benchmark
14
+ - parallel-corpus
15
+ pretty_name: English-Serbian Semantic Text Similarity (STS) Benchmark
16
+ size_categories:
17
+ - n<1K
18
+ ---
19
+
20
+ # Dataset Card for English-Serbian Semantic Text Similarity Benchmark
21
+
22
+ ## Dataset Description
23
+
24
+ - **Repository:** [Provide a link to your dataset repository]
25
+
26
+ ### Dataset Summary
27
+
28
+ This dataset is a parallel English-Serbian Semantic Text Similarity (STS) benchmark. It was created to evaluate multilingual English-Serbian language models, with a focus on SBERT (Sentence-BERT) knowledge distillation. The dataset consists of sentence pairs in English and Serbian, along with their semantic similarity scores.
29
+ The dataset uses the test split from the original STS benchmark. Only the second sentence of each pair was translated to Serbian. The translation was done automatically using the `gpt-3.5-turbo-0125 model`. The prompt used for the translation can be found in the repository.
30
+
31
+ ### Supported Tasks and Leaderboards
32
+
33
+ - **Semantic Similarity:** The primary task for this dataset is to measure the semantic similarity between English-Serbian sentence pairs.
34
+ - **Cross-lingual Understanding:** This dataset can be used to evaluate models' performance in understanding semantic relationships across English and Serbian languages.
35
+ - **Machine Translation:** While not the primary focus, this dataset can also be used to assess English to Serbian translation quality.
36
+
37
+ ### Languages
38
+
39
+ The dataset contains text in two languages:
40
+ - English (en)
41
+ - Serbian (sr)
42
+
43
+ ## Dataset Structure
44
+
45
+ ### Data Instances
46
+
47
+ Each instance in the dataset consists of:'
48
+ 1. Datast split
49
+ 2. Genre (the domain the sentence belongs to)
50
+ 3. Dataset (original dataset a setence is obtained from)
51
+ 4. Year (which itteration of the STS dataset a sentence belogs to)
52
+ 5. Sid (a unique sentence id for each sentence pair)
53
+ 7. Sentence 1 (an original English sentence from the STS benchmark datast)
54
+ 8. Sentence 2 in English (an original English sentence from the STS benchmark datast)
55
+ 9. Setnece 2 in Serbian (an automatically translated Sentence 2 from the STS benchmark dataset)
56
+ 10. A similarity score (kept from the original STS benchmark dataset)
57
+
58
+ ### Data Fields
59
+
60
+ - `en_sentence`: The original English sentence
61
+ - `sr_sentence`: The corresponding Serbian translation
62
+ - `similarity_score`: A float value representing the semantic similarity between the two sentences
63
+
64
+ ### Data Splits
65
+
66
+ The dataset uses the test split from the original STS benchmark. Only the second sentence of each pair was translated to Serbian.
67
+
68
+ ## Dataset Creation
69
+
70
+ ### Curation Rationale
71
+
72
+ This dataset was created to address the need for multilingual semantic similarity benchmarks, specifically for the English-Serbian language pair. It aims to facilitate research in cross-lingual understanding and evaluation of multilingual language models.
73
+
74
+ ### Source Data
75
+
76
+ The source data used for the translation is the test split of the STS benchmark dataset from: https://huggingface.co/datasets/mteb/stsbenchmark-sts
77
+
78
+ #### Initial Data Collection and Normalization
79
+
80
+ The source data comes from the test split of the original STS benchmark dataset.
81
+
82
+ ## Considerations for Using the Data
83
+
84
+ ### Discussion of Biases
85
+
86
+ As the Serbian translations were produced automatically, there may be biases or errors introduced by the machine translation process. Users should be aware of potential gender, cultural, or linguistic biases that may be present in the original English sentences or introduced during translation.
87
+
88
+ ### Other Known Limitations
89
+
90
+ - The dataset is limited to the test split of the original STS benchmark, which may affect its size and diversity.
91
+ - The dataset may not cover all domains or types of semantic relationships.
92
+ - Automatic translation may not capture all nuances of the original English sentences. A random split of 100 translated sentences was validated. However, translation does not involve post-correction at this stage. Thus, the translated Serbian sentences might contain inconsistencies in relation to the rules and grammar of the standard Serbian language.
93
+
94
+ ### Licensing Information
95
+
96
+ This dataset is published under the MIT license.
97
+
98
+ ### Citation Information
99
+
100
+ If you use this dataset, please link the smartcat huggingface dataset repository.
101
+
102
+ ## Loading the Dataset
103
+
104
+ To load this dataset using the Hugging Face `datasets` library, you can use the following code:
105
+
106
+ ```python
107
+ from datasets import load_dataset
108
+
109
+ # Replace 'username/dataset_name' with the actual path to your dataset on Hugging Face
110
+ dataset = load_dataset("smartcat/STS_parallel_en_sr")
111
+
112
+ # Access the data
113
+ for example in dataset['train']:
114
+ print(f"Sentence 1: {example[sentence1]}")
115
+ print(f"Sentence 2 in English: {example['sentence2_eng']}")
116
+ print(f"Sentence 2 in Serbian: {example['sentence2_sr']}")
117
+ print(f"Similarity Score: {example['score']}")
118
+ print()