Upload AirQA.py with huggingface_hub

AirQA.py

@@ -0,0 +1,62 @@
+import json
+import datasets
+
+class AirQA(datasets.GeneratorBasedBuilder):
+    """Example dataset loader for JSONL with selective fields and only test split."""
+
+    VERSION = datasets.Version("1.0.0")
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description="AirQA: A Comprehensive QA Dataset for AI Research with Instance-Level Evaluation",
+            features=datasets.Features({
+                "uuid": datasets.Value("string"),
+                "question": datasets.Value("string"),
+                "answer_format": datasets.Value("string"),
+                "anchor_pdf": datasets.Sequence(datasets.Value("string")),
+                "reference_pdf": datasets.Sequence(datasets.Value("string")),
+                "conference": datasets.Sequence(datasets.Value("string")),
+            }),
+            supervised_keys=None,
+            homepage="https://huggingface.co/datasets/OpenDFM/AirQA",
+            license="apache-2.0",
+        )
+
+    def _split_generators(self, dl_manager):
+        files = dl_manager.download_and_extract({
+            "data": "data/test_data.jsonl",
+            "map": "data/uuid2title.json",
+        })
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TEST,
+                gen_kwargs={
+                    "path": files["data"],
+                    "map_path": files["map"],
+                },
+            )
+        ]
+
+    def _generate_examples(self, path, map_path):
+        with open(map_path, encoding="utf-8") as f:
+            uuid2title = json.load(f)
+
+        keep_keys = ["uuid", "question", "answer_format", "anchor_pdf", "reference_pdf", "conference"]
+
+        def map_list(lst):
+            if lst is None:
+                return None
+            return [uuid2title.get(x, x) for x in lst]
+
+        with open(path, encoding="utf-8") as f:
+            for i, line in enumerate(f):
+                if not line.strip():
+                    continue
+                ex = json.loads(line)
+
+                filtered = {k: ex.get(k, None) for k in keep_keys}
+
+                filtered["anchor_pdf"] = map_list(filtered["anchor_pdf"])
+                filtered["reference_pdf"] = map_list(filtered["reference_pdf"])
+
+                yield i, filtered