This pull request makes the following improvements to the DeepScholarBench dataset card:
* **Metadata**:
* Updates the `task_categories` metadata field to `text-generation` to accurately reflect the dataset's primary purpose as a benchmark for generative research synthesis, as described in the paper.
* Adds several descriptive `tags` like `scientific-research`, `academic-papers`, `citation-analysis`, `retrieval-augmented-generation`, `rag`, `summarization`, and `llm-evaluation` to improve discoverability and provide a comprehensive overview of the dataset's applications.
* **Content**:
* Adds a link to the official Hugging Face paper page (`https://huggingface.co/papers/2508.20033`) alongside the existing ArXiv link, enhancing discoverability on the Hub. The existing paper badge has been clarified as `Paper (arXiv)`.
* Includes the paper's abstract in a dedicated "
