Dataset Viewer
Auto-converted to Parquet Duplicate
query-id
stringlengths
1
6
corpus-id
stringlengths
2
6
score
stringclasses
1 value
52462
49462
1
63024
68660
1
63022
62860
1
10703
10619
1
10703
73282
1
10703
90412
1
10703
10701
1
21616
52365
1
46862
65485
1
46866
46762
1
44426
3127
1
97939
80376
1
12833
111676
1
23129
10788
1
23129
107228
1
50415
25518
1
52818
104162
1
60905
57067
1
75289
14542
1
99850
24442
1
21937
21033
1
99993
61243
1
112238
25494
1
33756
40674
1
102106
37787
1
109853
30084
1
109853
30106
1
103703
88092
1
72392
112378
1
72393
26892
1
103709
9020
1
101706
88865
1
92748
58680
1
45449
87329
1
45449
63046
1
45449
59026
1
45449
81103
1
45449
103494
1
45449
102779
1
45449
103359
1
8032
40718
1
8032
19494
1
8032
17679
1
88723
91977
1
89232
78510
1
56843
17712
1
19580
75376
1
53569
66194
1
62121
91638
1
28406
25936
1
110551
4376
1
55129
54291
1
23717
45405
1
59203
57642
1
78923
77018
1
104165
104161
1
57784
36801
1
75092
113743
1
57438
2555
1
36774
28353
1
27822
88053
1
10352
88806
1
70301
68359
1
49844
31706
1
49844
52750
1
49844
52250
1
48872
53676
1
82175
73836
1
94280
80589
1
29257
20331
1
72051
72242
1
72051
72176
1
68843
104142
1
1449
58647
1
1449
4346
1
35624
14323
1
35624
46701
1
35624
19135
1
35624
25727
1
35624
19098
1
35624
33423
1
25522
8348
1
74244
9662
1
41829
85296
1
75119
75296
1
86444
87000
1
86444
86773
1
14142
7839
1
14142
111490
1
14142
17537
1
14142
10756
1
14142
113435
1
87478
84572
1
6470
80
1
6470
97872
1
110409
110372
1
46716
48961
1
46716
75485
1
68956
69577
1
38855
21761
1
End of preview. Expand in Data Studio

Dataset Summary

CQADupstack-gis-Fa is a Persian (Farsi) dataset developed for the Retrieval task, with a focus on duplicate question detection in community question-answering (CQA) platforms. This dataset is a translated version of the "GIS" (Geographic Information Systems) StackExchange subforum from the English CQADupstack collection and is part of the FaMTEB benchmark under the BEIR-Fa suite.

  • Language(s): Persian (Farsi)
  • Task(s): Retrieval (Duplicate Question Retrieval)
  • Source: Translated from English using Google Translate
  • Part of FaMTEB: Yes — under BEIR-Fa

Supported Tasks and Leaderboards

This dataset evaluates text embedding models on their ability to retrieve semantically equivalent or duplicate questions in the GIS domain. Results can be viewed on the Persian MTEB Leaderboard (filter by language: Persian).

Construction

The dataset was created by:

  • Translating the "gis" subforum from the English CQADupstack dataset using the Google Translate API
  • Preserving the CQA structure for evaluating duplicate retrieval tasks

According to the FaMTEB paper, the BEIR-Fa datasets underwent translation quality evaluation using:

  • BM25 retrieval performance comparisons (English vs Persian)
  • GEMBA-DA evaluation, leveraging LLMs for quality control

Data Splits

As per the FaMTEB paper (Table 5), all CQADupstack-Fa sub-datasets share a pooled evaluation set:

  • Train: 0 samples
  • Dev: 0 samples
  • Test: 480,902 samples (aggregate)

The cqadupstack-gis-fa subset contains approximately 39.6k examples (user-provided figure). For more granular splits, consult the dataset provider or FaMTEB documentation.

Downloads last month
21