--- license: apache-2.0 task_categories: - text-classification - zero-shot-classification pretty_name: arxiv-metadata-oai-snapshot size_categories: - 10K<n<100K --- # arXiv Computer Science Dataset Description ## Overview This dataset is a filtered subset of arXiv papers focusing exclusively on Computer Science domains. It contains 46,328 academic papers with their metadata and classifications. ## Dataset Structure The dataset contains 7 columns: 1. `id`: Unique identifier for each arXiv paper 2. `submitter`: The individual who submitted the paper to arXiv 3. `title`: The title of the research paper 4. `categories`: Original arXiv category codes (e.g., cs.AI, cs.CL) 5. `abstract`: The paper's abstract text 6. `labels`: Transformed category names in human-readable format 7. `domain`: The primary domain classification (Computer Science) ## Data Processing - The dataset was created by filtering the complete arXiv metadata dump for Computer Science papers - Category codes were mapped to their full names using a category taxonomy - Each paper's primary domain was extracted from its categories - Only papers with at least one Computer Science category were included ## Data Format - File format: CSV - Number of records: 46,328 ## Usage This dataset is suitable for: - Text classification tasks - Topic modeling - Research trend analysis - Bibliometric studies - Computer Science subdomain classification ## Source The data was derived from the arXiv metadata snapshot and processed using a custom category taxonomy mapping.