Add dataset

README.md

@@ -0,0 +1,12 @@
+---
+language: en
+license: apache-2.0
+---
+
+# Dataset card for Wikipedia Domain Labels
+
+This dataset is a list of domain labels for the [txtai-wikipedia-slim](https://huggingface.co/NeuML/txtai-wikipedia-slim) embeddings database, which is the Top 100K most viewed Wikipedia articles.
+
+## Training code
+
+[The training code used to build this model is here](https://huggingface.co/datasets/NeuML/wikipedia-domain-labels/blob/main/labels.py).

labels.py

@@ -0,0 +1,46 @@
+import csv
+
+from tqdm.auto import tqdm
+from txtai import Embeddings
+from txtai.pipeline import Labels
+
+LABELS = [
+    "aerospace", "agronomy", "artistic", "astronomy", "atmospheric_science", "automotive", "beauty",
+    "biology", "celebrity", "chemistry", "civil_engineering", "communication_engineering",
+    "computer_science_and_technology", "design", "drama_and_film", "economics",
+    "electronic_science", "entertainment", "environmental_science", "fashion", "finance",
+    "food", "gamble", "game", "geography", "health", "history", "hobby", "hydraulic_engineering", 
+    "instrument_science", "journalism_and_media_communication", "landscape_architecture", "law",
+    "library", "literature", "materials_science", "mathematics", "mechanical_engineering",
+    "medical", "mining_engineering", "movie", "music_and_dance", "news", "nuclear_science",
+    "ocean_science", "optical_engineering", "painting", "pet", 
+    "petroleum_and_natural_gas_engineering", "philosophy", "photo", "physics", "politics",
+    "psychology", "public_administration", "relationship", "religion", "sociology", "sports",
+    "statistics", "systems_science", "textile_science", "topicality", "transportation_engineering",
+    "travel", "urban_planning", "vulgar_language",
+]
+
+
+def stream():
+    for row in tqdm(rows):
+        yield row["text"]
+
+
+labels = Labels("MoritzLaurer/deberta-v3-large-zeroshot-v2.0-c")
+
+embeddings = Embeddings()
+embeddings.load(provider="huggingface-hub", container="neuml/txtai-wikipedia-slim")
+
+rows = embeddings.search("SELECT id, text FROM txtai", 100_000)
+
+print("NUMBER OF LABELS:", len(LABELS))
+
+with open("labels.csv", "w", encoding="utf-8", newline="") as f:
+    writer = csv.writer(f)
+
+    # Write header
+    writer.writerow(["id", "label"])
+
+    # Write each row one by one
+    for x, label in enumerate(labels(stream(), LABELS, flatten=True, batch_size=32)):
+        writer.writerow([rows[x]["id"], label[0]])