fix: Add Korean AutoRAGRetrieval (#1388)

* feat: add AutoRAG Korean embedding retrieval benchmark

* fix: run --- 🧹 Running linters ---
ruff format . 			# running ruff formatting
716 files left unchanged
ruff check . --fix  	# running ruff linting
All checks passed!

* fix: add metadata for AutoRAGRetrieval

* change link for markers_bm

* add AutoRAGRetrieval to init.py and update metadata

* add precise metadata

* update metadata: description and license

* delete descriptive_stats in AutoRAGRetrieval.py and run calculate_matadata_metrics.py

mteb/descriptive_stats/Retrieval/AutoRAGRetrieval.json

@@ -0,0 +1,11 @@
+{
+    "test": {
+        "number_of_characters": 894.2168128654971,
+        "num_samples": 834,
+        "num_queries": 114,
+        "num_documents": 720,
+        "average_document_length": 1.1452816358024691,
+        "average_query_length": 0.610649430594029,
+        "average_relevant_docs_per_query": 1.0
+    }
+}

mteb/tasks/Retrieval/__init__.py

@@ -100,6 +100,7 @@
 from .jpn.NLPJournalTitleAbsRetrieval import *
 from .jpn.NLPJournalTitleIntroRetrieval import *
 from .kat.GeorgianFAQRetrieval import *
+from .kor.AutoRAGRetrieval import *
 from .kor.KoStrategyQA import *
 from .multilingual.BelebeleRetrieval import *
 from .multilingual.CrossLingualSemanticDiscriminationWMT19 import *

mteb/tasks/Retrieval/kor/AutoRAGRetrieval.py

@@ -0,0 +1,40 @@
+from __future__ import annotations
+
+from mteb.abstasks.TaskMetadata import TaskMetadata
+
+from ....abstasks.AbsTaskRetrieval import AbsTaskRetrieval
+
+
+class AutoRAGRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="AutoRAGRetrieval",
+        description="This dataset enables the evaluation of Korean RAG performance across various domains—finance, public sector, healthcare, legal, and commerce—by providing publicly accessible documents, questions, and answers.",
+        reference="https://arxiv.org/abs/2410.20878",
+        dataset={
+            "path": "yjoonjang/markers_bm",
+            "revision": "fd7df84ac089bbec763b1c6bb1b56e985df5cc5c",
+        },
+        type="Retrieval",
+        prompt="Retrieve text based on user query.",
+        category="s2p",
+        modalities=["text"],
+        eval_splits=["test"],
+        eval_langs=["kor-Hang"],
+        main_score="ndcg_at_10",
+        date=("2024-08-03", "2024-08-03"),
+        domains=["Government", "Medical", "Legal", "Social"],
+        task_subtypes=["Article retrieval"],
+        license="mit",
+        annotations_creators="human-annotated",
+        dialect=[],
+        sample_creation="created",
+        bibtex_citation="""@misc{kim2024autoragautomatedframeworkoptimization,
+      title={AutoRAG: Automated Framework for optimization of Retrieval Augmented Generation Pipeline}, 
+      author={Dongkyu Kim and Byoungwook Kim and Donggeon Han and Matouš Eibich},
+      year={2024},
+      eprint={2410.20878},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL},
+      url={https://arxiv.org/abs/2410.20878}, 
+}""",
+    )