Add metadata and improve dataset card

README.md

@@ -1,19 +1,46 @@
 ---
 dataset_info:
   features:
-    # 1. 把你想先看到的字段放在最前面，顺序自定义
-    - name: language
-      dtype: string
-    - name: article
-      dtype: string
-    - name: summary
-      dtype: string
-    - name: output
-      dtype: string
-    - name: instruction
-      dtype: string
-    - name: input
-      dtype: string
-    - name: id
-      dtype: string
----
+  - name: language
+    dtype: string
+  - name: article
+    dtype: string
+  - name: summary
+    dtype: string
+  - name: output
+    dtype: string
+  - name: instruction
+    dtype: string
+  - name: input
+    dtype: string
+  - name: id
+    dtype: string
+task_categories:
+- text-classification
+language:
+- zh
+- en
+license: apache-2.0
+---
+
+# LongNovel: A Multi-Scale Benchmark for Hallucination Detection in Long-Context Novel Summarization
+
+LongNovel is a multi-scale, bilingual (Chinese and English) benchmark for hallucination detection in long-context novel summarization. It is constructed from 29 Chinese novels (16k–100k tokens) and chapter-level data from the BookSum dataset, and defines 8 distinct hallucination types.
+
+Paper: [LongNovel: A Multi-Scale Benchmark for Hallucination Detection in Long-Context Novel Summarization](https://huggingface.co/papers/2608.18082)  
+Code: [https://github.com/BDML-lab/LongNovel](https://github.com/BDML-lab/LongNovel)
+
+## Usage
+
+```python
+from datasets import load_dataset
+
+# Load the dataset
+dataset = load_dataset('SII-BDML/LongNovel', split='test')
+```
+
+Each instance contains the fields: `id`, `language`, `article`, `summary`, `instruction`, `input`, and `output`.
+
+## License
+
+Apache License 2.0