Canonical pedometric vocabulary for LLM-KG claims

Controlled list of variable names used by the prompt template in causal_llm_extract() and by the gold-standard annotation tooling. Keeping these labels stable across thousands of abstracts is essential for computing precision / recall against the gold standard: free-form labels like "tree_cover" vs "tree cover" vs "vegetation_cover" would artificially depress matching accuracy.

Usage

llm_annotation_vocabulary()

Value

Character vector of canonical variable names.

Examples

head(llm_annotation_vocabulary(), 10)
#>  [1] "precipitation"             "mean_annual_precipitation"
#>  [3] "temperature"               "mean_annual_temperature"  
#>  [5] "elevation"                 "slope"                    
#>  [7] "aspect"                    "twi"                      
#>  [9] "clay"                      "sand"