fix: Add JMTEB Clustering datasets (#768)

* Add JMTEB Clustering datasets
* Add points
* Add points for review

docs/mmteb/points/768.jsonl

@@ -0,0 +1,2 @@
+{"GitHub": "awinml", "New dataset": 8}
+{"GitHub": "isaac-chung", "Review PR": 2}

mteb/tasks/Clustering/__init__.py

@@ -25,6 +25,8 @@
 from .fra.AlloProfClusteringP2P import *
 from .fra.AlloProfClusteringS2S import *
 from .fra.HALClusteringS2S import *
+from .jpn.LivedoorNewsClustering import *
+from .jpn.MewsC16JaClustering import *
 from .multilingual.IndicReviewsClusteringP2P import *
 from .multilingual.MasakhaNEWSClusteringP2P import *
 from .multilingual.MasakhaNEWSClusteringS2S import *

mteb/tasks/Clustering/jpn/LivedoorNewsClustering.py

@@ -0,0 +1,37 @@
+from mteb.abstasks.AbsTaskClusteringFast import AbsTaskClusteringFast
+from mteb.abstasks.TaskMetadata import TaskMetadata
+
+
+class LivedoorNewsClustering(AbsTaskClusteringFast):
+    metadata = TaskMetadata(
+        name="LivedoorNewsClustering",
+        description="Clustering of the news reports of a Japanese news site, Livedoor News by RONDHUIT Co, Ltd. in 2012. It contains over 7,000 news report texts across 9 categories (topics).",
+        reference="https://github.com/sbintuitions/JMTEB",
+        dataset={
+            "path": "sbintuitions/JMTEB",
+            "name": "livedoor_news",
+            "revision": "e4af6c73182bebb41d94cb336846e5a452454ea7",
+        },
+        type="Clustering",
+        category="s2s",
+        eval_splits=["test"],
+        eval_langs=["jpn-Jpan"],
+        main_score="v_measure",
+        date=("2000-01-01", "2014-02-09"),
+        form=["written"],
+        domains=["News"],
+        task_subtypes=["Topic classification"],
+        license="cc-by-nd-2.1-jp",
+        socioeconomic_status="high",
+        annotations_creators="derived",
+        dialect=[],
+        text_creation="found",
+        bibtex_citation="",
+        n_samples={"test": 1107},
+        avg_character_length={"test": 1082.61},
+    )
+
+    def dataset_transform(self):
+        self.dataset = self.dataset.rename_columns(
+            {"text": "sentences", "label": "labels"}
+        )

mteb/tasks/Clustering/jpn/MewsC16JaClustering.py

@@ -0,0 +1,57 @@
+from mteb.abstasks.AbsTaskClusteringFast import AbsTaskClusteringFast
+from mteb.abstasks.TaskMetadata import TaskMetadata
+
+
+class MewsC16JaClustering(AbsTaskClusteringFast):
+    metadata = TaskMetadata(
+        name="MewsC16JaClustering",
+        description="""MewsC-16 (Multilingual Short Text Clustering Dataset for News in 16 languages) is constructed from Wikinews.
+        This dataset is the Japanese split of MewsC-16, containing topic sentences from Wikinews articles in 12 categories.
+        More detailed information is available in the Appendix E of the citation.""",
+        reference="https://github.com/sbintuitions/JMTEB",
+        dataset={
+            "path": "sbintuitions/JMTEB",
+            "name": "mewsc16_ja",
+            "revision": "e4af6c73182bebb41d94cb336846e5a452454ea7",
+        },
+        type="Clustering",
+        category="s2s",
+        eval_splits=["test"],
+        eval_langs=["jpn-Jpan"],
+        main_score="v_measure",
+        date=("2002-01-01", "2019-01-31"),
+        form=["written"],
+        domains=["News"],
+        task_subtypes=["Topic classification"],
+        license="cc-by-sa-4.0",
+        socioeconomic_status="high",
+        annotations_creators="derived",
+        dialect=[],
+        text_creation="found",
+        bibtex_citation="""
+        @inproceedings{
+            nishikawa-etal-2022-ease,
+            title = "{EASE}: Entity-Aware Contrastive Learning of Sentence Embedding",
+            author = "Nishikawa, Sosuke  and
+            Ri, Ryokan  and
+            Yamada, Ikuya  and
+            Tsuruoka, Yoshimasa  and
+            Echizen, Isao",
+            booktitle = "Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
+            month = jul,
+            year = "2022",
+            address = "Seattle, United States",
+            publisher = "Association for Computational Linguistics",
+            url = "https://aclanthology.org/2022.naacl-main.284",
+            pages = "3870--3885",
+            abstract = "We present EASE, a novel method for learning sentence embeddings via contrastive learning between sentences and their related entities.The advantage of using entity supervision is twofold: (1) entities have been shown to be a strong indicator of text semantics and thus should provide rich training signals for sentence embeddings; (2) entities are defined independently of languages and thus offer useful cross-lingual alignment supervision.We evaluate EASE against other unsupervised models both in monolingual and multilingual settings.We show that EASE exhibits competitive or better performance in English semantic textual similarity (STS) and short text clustering (STC) tasks and it significantly outperforms baseline methods in multilingual settings on a variety of tasks.Our source code, pre-trained models, and newly constructed multi-lingual STC dataset are available at https://github.com/studio-ousia/ease.",
+        }
+        """,
+        n_samples={"test": 992},
+        avg_character_length={"test": 95},
+    )
+
+    def dataset_transform(self):
+        self.dataset = self.dataset.rename_columns(
+            {"text": "sentences", "label": "labels"}
+        )

results/intfloat__multilingual-e5-small/LivedoorNewsClustering.json

@@ -0,0 +1,23 @@
+{
+  "dataset_revision": "e4af6c73182bebb41d94cb336846e5a452454ea7",
+  "mteb_dataset_name": "LivedoorNewsClustering",
+  "mteb_version": "1.10.15",
+  "test": {
+    "evaluation_time": 467.76,
+    "v_measure": 0.539826287544948,
+    "v_measures": {
+      "Level 0": [
+        0.5375342521798784,
+        0.5910943502833345,
+        0.6145159457550844,
+        0.5405749178389159,
+        0.5171663369190832,
+        0.5002664592267217,
+        0.5510727077832276,
+        0.4776535022956716,
+        0.5564596669393072,
+        0.5119247362282552
+      ]
+    }
+  }
+}

results/intfloat__multilingual-e5-small/MewsC16JaClustering.json

@@ -0,0 +1,23 @@
+{
+  "dataset_revision": "e4af6c73182bebb41d94cb336846e5a452454ea7",
+  "mteb_dataset_name": "MewsC16JaClustering",
+  "mteb_version": "1.10.15",
+  "test": {
+    "evaluation_time": 66.46,
+    "v_measure": 0.445783961624162,
+    "v_measures": {
+      "Level 0": [
+        0.45909442255508964,
+        0.46305961184346073,
+        0.4427483337819293,
+        0.47263199422184127,
+        0.47254962635535197,
+        0.42725383372955317,
+        0.43294497964596884,
+        0.44843650657068573,
+        0.4320901331069621,
+        0.40703017443077744
+      ]
+    }
+  }
+}

results/sentence-transformers__paraphrase-multilingual-MiniLM-L12-v2/LivedoorNewsClustering.json

@@ -0,0 +1,23 @@
+{
+  "dataset_revision": "e4af6c73182bebb41d94cb336846e5a452454ea7",
+  "mteb_dataset_name": "LivedoorNewsClustering",
+  "mteb_version": "1.10.15",
+  "test": {
+    "evaluation_time": 134.32,
+    "v_measure": 0.37346528729902917,
+    "v_measures": {
+      "Level 0": [
+        0.34107725561094543,
+        0.3791343290309538,
+        0.3416715311884291,
+        0.42353515150692833,
+        0.38305753322534697,
+        0.3330157742849174,
+        0.37443106402827003,
+        0.3557980434687236,
+        0.41328801563860723,
+        0.38964417500716964
+      ]
+    }
+  }
+}

results/sentence-transformers__paraphrase-multilingual-MiniLM-L12-v2/MewsC16JaClustering.json

@@ -0,0 +1,23 @@
+{
+  "dataset_revision": "e4af6c73182bebb41d94cb336846e5a452454ea7",
+  "mteb_dataset_name": "MewsC16JaClustering",
+  "mteb_version": "1.10.15",
+  "test": {
+    "evaluation_time": 62.81,
+    "v_measure": 0.4343534234258306,
+    "v_measures": {
+      "Level 0": [
+        0.46723817526632816,
+        0.4502995883519553,
+        0.4258278925500268,
+        0.4352989001355879,
+        0.41773899563411776,
+        0.42619246704315167,
+        0.41465632852404943,
+        0.4116361040552486,
+        0.4774990451040913,
+        0.4171467375937491
+      ]
+    }
+  }
+}