added scidcc

climabench.py

@@ -23,7 +23,7 @@ import json
 
 
 _CITATION = """
-@misc{laud2023climabench,
+@misc{laud2023Climabench,
       title={ClimaBench: A Benchmark Dataset For Climate Change Text Understanding in English}, 
       author={Tanmay Laud and Daniel Spokoyny and Tom Corringham and Taylor Berg-Kirkpatrick},
       year={2023},
@@ -37,7 +37,7 @@ _DESCRIPTION = """\
 The topic of Climate Change (CC) has received limited attention in NLP despite its real world urgency.
 Activists and policy-makers need NLP tools in order to effectively process the vast and rapidly growing textual data produced on CC.
 Their utility, however, primarily depends on whether the current state-of-the-art models can generalize across various tasks in the CC domain.
-In order to address this gap, we introduce Climate Change Benchmark (ClimaBench), a benchmark collection of existing disparate datasets for evaluating model performance across a diverse set of CC NLU tasks systematically.
+In order to address this gap, we introduce Climate Change Benchmark (Climabench), a benchmark collection of existing disparate datasets for evaluating model performance across a diverse set of CC NLU tasks systematically.
 Further, we enhance the benchmark by releasing two large-scale labelled text classification and question-answering datasets curated from publicly available environmental disclosures.
 Lastly, we provide an analysis of several generic and CC-oriented models answering whether fine-tuning on domain text offers any improvements across these tasks. We hope this work provides a standard assessment tool for research on CC text data.
 """
@@ -47,15 +47,15 @@ _HOMEPAGE = "https://arxiv.org/abs/2301.04253"
 _LICENSE = ""
 
 _URL = "https://drive.google.com/u/0/uc?id=1RnZDC-KMOx8JkhbrFl86TsDpsskZB6fO&export=download&confirm=t&uuid=33b5ea55-a99a-431c-b805-0f298ed2912a&at=AKKF8vzyC_uPCdXvV4-gPlJ11X1J:1688024195786"
-
-_CLIMABENCH_BASE_KWARGS = dict(
+# _URL = "https://drive.google.com/drive/folders/1zaOPdS88m2oMRqmDGs4KqtWEIRJcPsMO"
+_Climabench_BASE_KWARGS = dict(
     citation=_CITATION,
     url=_HOMEPAGE,
 )
 
 
-class ClimaBenchConfig(datasets.BuilderConfig):
-    """BuilderConfig for ClimaBench."""
+class ClimabenchConfig(datasets.BuilderConfig):
+    """BuilderConfig for Climabench."""
 
     def __init__(
         self,
@@ -68,7 +68,7 @@ class ClimaBenchConfig(datasets.BuilderConfig):
         process_label=lambda x: x,
         **kwargs,
     ):
-        """BuilderConfig for ClimaBench.
+        """BuilderConfig for Climabench.
         Args:
           text_features: `dict[string, string]`, map from the name of the feature
             dict for each text field to the name of the column in the tsv file
@@ -85,7 +85,7 @@ class ClimaBenchConfig(datasets.BuilderConfig):
             of the label and processing it to the form required by the label feature
           **kwargs: keyword arguments forwarded to super.
         """
-        super(ClimaBenchConfig, self).__init__(
+        super(ClimabenchConfig, self).__init__(
             version=datasets.Version("1.0.0", ""), **kwargs
         )
         self.text_features = text_features
@@ -97,11 +97,11 @@ class ClimaBenchConfig(datasets.BuilderConfig):
         self.process_label = process_label
 
 
-class ClimaBench(datasets.GeneratorBasedBuilder):
+class Climabench(datasets.GeneratorBasedBuilder):
     """FLORES-200 dataset."""
 
     BUILDER_CONFIGS = [
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="climate_stance",
             description=textwrap.dedent(
                 """\
@@ -132,7 +132,7 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             ),
             url="https://github.com/roopalv54/finegrained-climate-change-social-media",
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="climate_eng",
             description=textwrap.dedent(
                 """\
@@ -163,7 +163,7 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             ),
             url="https://github.com/roopalv54/finegrained-climate-change-social-media",
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="climate_fever",
             description=textwrap.dedent(
                 """\
@@ -184,7 +184,7 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             ),
             url="http://climatefever.ai",
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="climatext",
             description=textwrap.dedent(
                 """\
@@ -208,7 +208,7 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             ),
             url="https://www.sustainablefinance.uzh.ch/en/research/climate-fever/climatext.html",
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="clima_insurance",
             description=textwrap.dedent(
                 """\
@@ -222,9 +222,9 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             data_dir="all_data/ClimateInsurance",
             text_features={"text": "text"},
             label_column="label",
-            **_CLIMABENCH_BASE_KWARGS,
+            **_Climabench_BASE_KWARGS,
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="clima_insurance_plus",
             description=textwrap.dedent(
                 """\
@@ -238,9 +238,9 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             data_dir="all_data/ClimateInsuranceMulti",
             text_features={"text": "text"},
             label_column="label",
-            **_CLIMABENCH_BASE_KWARGS,
+            **_Climabench_BASE_KWARGS,
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="clima_cdp",
             description=textwrap.dedent(
                 """\
@@ -271,9 +271,9 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             data_dir="all_data/CDP/Cities/Cities Responses",
             text_features={"text": "Text"},
             label_column="Label",
-            **_CLIMABENCH_BASE_KWARGS,
+            **_Climabench_BASE_KWARGS,
         ),
-        ClimaBenchConfig(
+        ClimabenchConfig(
             name="clima_qa",
             description=textwrap.dedent(
                 """\
@@ -291,7 +291,62 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
             text_features={"question": "question", "answer": "answer"},
             label_classes=["0", "1"],
             label_column="label",
-            **_CLIMABENCH_BASE_KWARGS,
+            **_Climabench_BASE_KWARGS,
+        ),
+        ClimabenchConfig(
+            name="scidcc",
+            description=textwrap.dedent(
+                """\
+            The Science Daily Climate Change (SCIDCC) dataset is
+            curated by web scraping news articles from the
+            Science Daily (SD) website. It contains around
+            11k news articles from 20 categories relevant to
+            climate change, where each article comprises of
+            a title, summary, and a body. Some of the major
+            categories are Earthquakes, Pollution, Hurricanes
+            and Cyclones. We propose to use this dataset for
+            the first time as a category classification task for
+            two reasons. Firstly, the SD news articles are relatively
+            more scientific as compared to other online
+            news. Secondly, the average document length is
+            around 500-600 words with a maximum of roughly
+            2.5k words, which is significantly longer than other
+            existing public CC datasets."""
+            ),
+            label_classes=[
+                "Ozone Holes",
+                "Pollution",
+                "Hurricanes Cyclones",
+                "Earthquakes",
+                "Climate",
+                "Environment",
+                "Geography",
+                "Geology",
+                "Global Warming",
+                "Weather",
+                "Agriculture & Food",
+                "Animals",
+                "Biology",
+                "Endangered Animals",
+                "Extinction",
+                "New Species",
+                "Zoology",
+                "Biotechnology",
+                "Genetically Modified",
+                "Microbes",
+            ],
+            data_dir="all_data",
+            text_features={"title": "Title", "summary": "Summary", "body": "Body"},
+            label_column="Category",
+            citation=textwrap.dedent(
+                """@inproceedings{mishra2021neuralnere,
+                title={NeuralNERE: Neural Named Entity Relationship Extraction for End-to-End Climate Change Knowledge Graph Construction},
+                author={Mishra, Prakamya and Mittal, Rohan},
+                booktitle={ICML 2021 Workshop on Tackling Climate Change with Machine Learning},
+                url={https://www.climatechange.ai/papers/icml2021/76},
+                year={2021}
+                }"""
+            ),
         ),
     ]
 
@@ -348,6 +403,17 @@ class ClimaBench(datasets.GeneratorBasedBuilder):
         dl_dir = dl_manager.download_and_extract(_URL)
         data_dir = os.path.join(dl_dir, self.config.data_dir)
 
+        if self.config.name == "scidcc":
+            return [
+                datasets.SplitGenerator(
+                    name=datasets.Split.TEST,
+                    gen_kwargs={
+                        "data_file": os.path.join(data_dir or "", "SciDCC.csv"),
+                        "split": "test",
+                    },
+                ),
+            ]
+
         if self.config.name == "climate_fever":
             return [
                 datasets.SplitGenerator(