query-id stringlengths 1 4 | corpus-id stringlengths 2 71 | score stringclasses 1 value |
|---|---|---|
0 | Habitat_destruction | 1 |
0 | Polar_bear | 1 |
0 | Extinction_risk_from_global_warming | 1 |
0 | Global_warming | 1 |
5 | Famine | 1 |
5 | Winter | 1 |
5 | Weather | 1 |
6 | Polar_bear | 1 |
9 | Atmosphere_of_Mars | 1 |
9 | Carbon_dioxide | 1 |
9 | Carbon_dioxide_in_Earth's_atmosphere | 1 |
10 | Ocean_acidification | 1 |
10 | Sea | 1 |
10 | Carbon_dioxide_in_Earth's_atmosphere | 1 |
11 | Carbon_dioxide | 1 |
11 | Greenhouse_gas | 1 |
11 | Petroleum | 1 |
14 | Great_Barrier_Reef | 1 |
14 | Coral_bleaching | 1 |
18 | Air_pollution | 1 |
18 | Global_catastrophic_risk | 1 |
19 | Terraforming_of_Mars | 1 |
19 | Carbon_dioxide | 1 |
19 | Carbon_dioxide_in_Earth's_atmosphere | 1 |
21 | Russia | 1 |
21 | Sea_level_rise | 1 |
21 | Sweden | 1 |
27 | Earth | 1 |
27 | Little_Ice_Age | 1 |
28 | Snowball_Earth | 1 |
28 | Climate_variability | 1 |
28 | Parinacota_(volcano) | 1 |
28 | Antarctica | 1 |
30 | Bushfires_in_Australia | 1 |
30 | 2007_Kangaroo_Island_bushfires | 1 |
30 | Natural_disaster | 1 |
31 | Carbon_tax | 1 |
31 | Permian–Triassic_extinction_event | 1 |
31 | Paleocene–Eocene_Thermal_Maximum | 1 |
33 | Fossil_fuel | 1 |
33 | Wind_power | 1 |
33 | Wind_turbine | 1 |
33 | Petroleum | 1 |
35 | Antarctic_ice_sheet | 1 |
35 | Sea_level_rise | 1 |
36 | Climate_change_denial | 1 |
36 | Global_cooling | 1 |
36 | Richard_Lindzen | 1 |
38 | Ice_age | 1 |
38 | Sea_level_rise | 1 |
38 | Antarctica | 1 |
41 | Scientific_consensus_on_climate_change | 1 |
41 | Sustainability | 1 |
42 | Sea_level | 1 |
42 | Sea_level_rise | 1 |
42 | Global_warming | 1 |
44 | Scientific_consensus_on_climate_change | 1 |
44 | Global_warming_controversy | 1 |
51 | Scientific_consensus_on_climate_change | 1 |
51 | Carbon_dioxide | 1 |
51 | Climate_change_adaptation | 1 |
51 | Greenhouse_gas | 1 |
55 | Hockey_stick_controversy | 1 |
57 | Medieval_Warm_Period | 1 |
60 | Scientific_consensus_on_climate_change | 1 |
60 | Patrick_Michaels | 1 |
60 | Global_warming | 1 |
61 | Space_Shuttle_Challenger_disaster | 1 |
61 | NASA | 1 |
61 | Global_warming | 1 |
65 | Windmill | 1 |
65 | Wind_turbine | 1 |
65 | Wind_farm | 1 |
67 | Earth | 1 |
67 | Holocene | 1 |
67 | Nordland | 1 |
67 | Beringia | 1 |
69 | Sea_level_rise | 1 |
71 | Sun | 1 |
71 | Solar_minimum | 1 |
71 | NASA | 1 |
72 | Begging_the_question | 1 |
72 | Narrative | 1 |
72 | Question_mark | 1 |
74 | Global_warming_controversy | 1 |
74 | Global_warming | 1 |
75 | Effects_of_global_warming | 1 |
75 | Scientific_consensus_on_climate_change | 1 |
75 | Weather | 1 |
76 | Earth | 1 |
76 | Climate_change_mitigation | 1 |
77 | Sun | 1 |
77 | Proxima_Centauri | 1 |
77 | Star | 1 |
79 | Scientific_consensus_on_climate_change | 1 |
79 | Intergovernmental_Panel_on_Climate_Change | 1 |
79 | IPCC_Fourth_Assessment_Report | 1 |
79 | Global_warming | 1 |
82 | Carbon_dioxide | 1 |
82 | Greenhouse_gas | 1 |
End of preview. Expand
in Data Studio
Dataset Summary
ClimateFEVER-Fa is a Persian (Farsi) dataset tailored for the Retrieval task, specifically focused on fact-checking climate-related claims. It is the translated version of the English ClimateFEVER dataset and is part of the FaMTEB (Farsi Massive Text Embedding Benchmark) under the BEIR-Fa collection.
- Language(s): Persian (Farsi)
- Task(s): Retrieval (Fact Checking, Evidence Retrieval for Climate Claims)
- Source: Translated from the English ClimateFEVER dataset
- Part of FaMTEB: Yes — under BEIR-Fa
Supported Tasks and Leaderboards
This dataset is designed to test how well text embedding models can retrieve relevant evidence from Persian Wikipedia to support or refute climate-related claims. Evaluation results can be found on the Persian MTEB Leaderboard (filter by language: Persian).
Construction
The dataset was constructed by:
- Translating the original English ClimateFEVER dataset into Persian using Google Translate
- Maintaining the original structure where each claim is used as a query, and the task is to retrieve relevant evidence passages from a Wikipedia corpus
As outlined in the FaMTEB paper:
- Translation quality was validated through BM25 score comparisons between English and Persian versions
- The GEMBA-DA framework was used with LLMs to directly assess the quality of translations, confirming satisfactory fidelity
Data Splits
Per FaMTEB paper (Table 5):
- Train: 0 samples
- Dev: 0 samples
- Test: 5,421,274 samples
Total: ~5.42 million examples
- Downloads last month
- 13