| --- |
| configs: |
| - config_name: default |
| data_files: |
| - split: test |
| path: qrels/test.jsonl |
| - config_name: corpus |
| data_files: |
| - split: corpus |
| path: corpus.jsonl |
| - config_name: queries |
| data_files: |
| - split: queries |
| path: queries.jsonl |
| --- |
| ## Dataset Summary |
|
|
| **ArguAna-Fa** is a Persian (Farsi) dataset designed for the **Retrieval** task, focusing on **argument and counter-argument retrieval**. It is a translated version of the original English **ArguAna** dataset used in the BEIR benchmark and is a part of the **FaMTEB** (Farsi Massive Text Embedding Benchmark) under the BEIR-Fa suite. |
|
|
| - **Language(s):** Persian (Farsi) |
| - **Task(s):** Retrieval (Argument Retrieval, Counter-Argument Retrieval) |
| - **Source:** Translated from the English ArguAna dataset using Google Translate |
| - **Part of FaMTEB:** Yes — part of the BEIR-Fa collection |
|
|
| ## Supported Tasks and Leaderboards |
|
|
| ArguAna-Fa is used to benchmark models on their ability to retrieve relevant **counterarguments** given an input argument. This tests the **semantic understanding of argumentation** in Persian. Performance can be evaluated on the **Persian MTEB Leaderboard** (filter by language: Persian). |
|
|
| ## Construction |
|
|
| - The dataset was created by translating the English ArguAna dataset using the **Google Translate API** |
| - Originally sourced from online debate portals, focusing on **argumentative reasoning and contrast** |
|
|
| As noted in the FaMTEB paper, the translation quality was evaluated by: |
|
|
| - Comparing **BM25 retrieval scores** between English and Persian |
| - Using the **GEMBA-DA framework** (LLM-based assessment) to ensure translation accuracy |
|
|
| ## Data Splits |
|
|
| According to the FaMTEB paper (Table 5): |
|
|
| - **Train:** 0 samples |
| - **Dev:** 0 samples |
| - **Test:** 10,080 samples |
|
|
| > Approximate total dataset size: **11.5k examples** (user-provided figure) |
|
|
