A collection of academic citation contexts categorized into three intent classes: Method, Background, and Result. Each record pairs a citation string with its corresponding section name and unique identifiers for both the citing and cited papers.
Use Cases
- Train a multi-class classifier to identify citation intent using the string text and label categories
- Investigate the relationship between paper structure and citation purpose using the sectionName and label columns
- Augment the dataset with external metadata like publication year or venue by querying the Semantic Scholar API with citingPaperId
- Develop natural language processing models to extract scientific claims from the string context field
Strengths
- Categorizes citations into three specific labels: 'Method', 'Background', and 'Result'
- Includes the sectionName field to provide structural context for the citation within the document
- Provides citingPaperId and citedPaperId for integration with the Semantic Scholar API
- Contains the string field which captures the specific textual context of the citation