query-id
stringlengths 1
5
| corpus-id
stringlengths 2
6
| score
stringclasses 1
value |
|---|---|---|
0
|
18850
|
1
|
4
|
196463
|
1
|
5
|
69306
|
1
|
6
|
560251
|
1
|
6
|
188530
|
1
|
6
|
564488
|
1
|
7
|
411063
|
1
|
9
|
509122
|
1
|
9
|
184698
|
1
|
11
|
596427
|
1
|
12
|
192516
|
1
|
12
|
338700
|
1
|
12
|
158738
|
1
|
13
|
503678
|
1
|
14
|
398960
|
1
|
16
|
60590
|
1
|
19
|
315086
|
1
|
19
|
142623
|
1
|
20
|
447231
|
1
|
21
|
497642
|
1
|
23
|
550624
|
1
|
23
|
32102
|
1
|
25
|
107584
|
1
|
25
|
562777
|
1
|
27
|
537326
|
1
|
28
|
250640
|
1
|
30
|
551175
|
1
|
30
|
434082
|
1
|
30
|
336922
|
1
|
30
|
19233
|
1
|
31
|
156554
|
1
|
32
|
279480
|
1
|
32
|
69623
|
1
|
32
|
84645
|
1
|
33
|
519798
|
1
|
33
|
425387
|
1
|
35
|
498681
|
1
|
35
|
80913
|
1
|
36
|
275249
|
1
|
36
|
368649
|
1
|
37
|
523564
|
1
|
41
|
176229
|
1
|
43
|
76662
|
1
|
45
|
284610
|
1
|
45
|
261220
|
1
|
46
|
91325
|
1
|
47
|
133299
|
1
|
48
|
108062
|
1
|
48
|
401260
|
1
|
48
|
329810
|
1
|
48
|
512151
|
1
|
49
|
352927
|
1
|
51
|
107817
|
1
|
51
|
257168
|
1
|
51
|
75195
|
1
|
52
|
566417
|
1
|
52
|
125111
|
1
|
53
|
84077
|
1
|
53
|
562798
|
1
|
53
|
102362
|
1
|
53
|
323269
|
1
|
53
|
119308
|
1
|
53
|
184852
|
1
|
53
|
119210
|
1
|
53
|
176196
|
1
|
54
|
590775
|
1
|
54
|
109546
|
1
|
54
|
511651
|
1
|
57
|
203633
|
1
|
57
|
391403
|
1
|
57
|
77818
|
1
|
57
|
226530
|
1
|
59
|
71601
|
1
|
61
|
524134
|
1
|
62
|
34810
|
1
|
63
|
509617
|
1
|
64
|
391619
|
1
|
65
|
580624
|
1
|
65
|
109203
|
1
|
66
|
397608
|
1
|
66
|
540395
|
1
|
66
|
405412
|
1
|
67
|
370815
|
1
|
67
|
294864
|
1
|
67
|
18792
|
1
|
67
|
511571
|
1
|
69
|
378484
|
1
|
70
|
327002
|
1
|
71
|
372052
|
1
|
72
|
549870
|
1
|
72
|
302049
|
1
|
74
|
46680
|
1
|
75
|
297965
|
1
|
76
|
375357
|
1
|
78
|
152407
|
1
|
80
|
252473
|
1
|
81
|
451207
|
1
|
82
|
500708
|
1
|
83
|
534277
|
1
|
85
|
431230
|
1
|
Dataset Summary
FiQA2018-Fa is a Persian (Farsi) dataset designed for the Retrieval task, specifically targeting opinion-based question answering in the financial domain. It is a translated version of the original English FiQA 2018 dataset and a core component of the FaMTEB (Farsi Massive Text Embedding Benchmark), under the BEIR-Fa collection.
- Language(s): Persian (Farsi)
- Task(s): Retrieval (Opinion-based Question Answering, Financial QA)
- Source: Translated from the English FiQA 2018 dataset using Google Translate
- Part of FaMTEB: Yes — under BEIR-Fa
Supported Tasks and Leaderboards
The dataset evaluates text embedding models on their ability to retrieve relevant financial content in response to subjective, opinion-based questions. Results are benchmarked on the Persian MTEB Leaderboard on Hugging Face Spaces (language filter: Persian).
Construction
Steps in dataset creation:
- Translation of the original English FiQA 2018 dataset (based on StackExchange "Investment" forum posts) using the Google Translate API
- The dataset retains mappings between user questions and relevant opinion-based answers
As outlined in the FaMTEB paper, the BEIR-Fa datasets (including FiQA2018-Fa) underwent:
- BM25 retrieval comparison with the original English
- Translation quality analysis using the GEMBA-DA LLM evaluation framework
These evaluations confirmed good translation quality for retrieval benchmarking.
Data Splits
According to the FaMTEB paper (Table 5):
- Train: 71,804 samples
- Dev: 0 samples
- Test: 59,344 samples
Total: ~131,148 examples
- Downloads last month
- 17