Datasets:
ArXiv:
License:
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -2,15 +2,16 @@
|
|
| 2 |
---
|
| 3 |
license: mit
|
| 4 |
---
|
| 5 |
-
# Normalized ConceptNet Database (Multi-Language,
|
| 6 |
|
| 7 |
This dataset contains a normalized version of the ConceptNet database.
|
| 8 |
-
**
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
Edges are kept if *at least one* node is in the target languages:
|
| 10 |
en, fr, it, de, es, ar, fa, grc, he, la, hbo.
|
| 11 |
|
| 12 |
Original data from: cstr/conceptnet-de-indexed
|
| 13 |
-
|
| 14 |
-
## Schema
|
| 15 |
-
... (schema is same as before) ...
|
| 16 |
|
|
|
|
| 2 |
---
|
| 3 |
license: mit
|
| 4 |
---
|
| 5 |
+
# Normalized ConceptNet Database (Multi-Language, V6)
|
| 6 |
|
| 7 |
This dataset contains a normalized version of the ConceptNet database.
|
| 8 |
+
**V6 FIX (Language Parsing):** This version correctly parses language codes
|
| 9 |
+
from ConceptNet URLs (e.g., `.../c/de/...` -> `de`) instead of
|
| 10 |
+
mistaking POS tags for languages. This fixes the data corruption
|
| 11 |
+
from V5.
|
| 12 |
+
|
| 13 |
Edges are kept if *at least one* node is in the target languages:
|
| 14 |
en, fr, it, de, es, ar, fa, grc, he, la, hbo.
|
| 15 |
|
| 16 |
Original data from: cstr/conceptnet-de-indexed
|
|
|
|
|
|
|
|
|
|
| 17 |
|