Recombination-Pred: Scientific Idea Recombination Edges from arXiv Abstracts

Recombination-Pred is a dataset for predicting how scientific concepts recombine, derived from the CHIMERA knowledge base. Each row represents a recombination edge between two concepts extracted from an arXiv abstract, with two defined types: combination and inspiration. The dataset was created by author 'noystl' and was last updated on May 30, 2026.

Use Cases

• Predicting novel scientific concept combinations based on recombination edges.
• Training models for knowledge graph link prediction using concept relationships.
• Analyzing patterns of idea inspiration across different scientific domains.
• Benchmarking algorithms for computational creativity and scientific discovery tasks.

Strengths

• Derived from the established CHIMERA knowledge base.
• Explicitly defines two recombination types: combination and inspiration.
• Data is sourced from scientific arXiv abstracts.
• Last updated on May 30, 2026, indicating recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

CHIMERA knowledge base and arXiv abstracts.

Collection Method

Likely extracted and processed from scientific paper abstracts.

Freshness

Last updated 2026-05-30 08:42:11