Compute Cosine Similarity Between Embeddings — foundry

Compute pairwise cosine similarity between all embeddings in a tibble. Useful for finding semantically similar texts.

Usage

foundry_similarity(data, text_col = "text", top_k = NULL, as_matrix = FALSE)

Arguments

data: A tibble from foundry_embed() containing an embedding list-column.
text_col: Character. Name of the column containing text labels. Default: "text".
top_k: Integer. Optional maximum number of most-similar pairs to return.
as_matrix: Logical. If TRUE, return the full cosine-similarity matrix instead of a long pairwise tibble.

Value

If as_matrix = FALSE, a tibble with columns:

text_1: Character. First text.
text_2: Character. Second text.
similarity: Numeric. Cosine similarity between -1 and 1.

If as_matrix = TRUE, a numeric cosine-similarity matrix with row and column names from text_col.

Examples

# Toy vectors demonstrate local computation without calling Azure.
embeddings <- tibble::tibble(
  text = c("Vector A", "Vector B", "Vector C"),
  embedding = list(c(1, 0), c(1, 1), c(0, 1))
)
foundry_similarity(embeddings)
#> # A tibble: 3 × 3
#>   text_1   text_2   similarity
#>   <chr>    <chr>         <dbl>
#> 1 Vector A Vector B      0.707
#> 2 Vector B Vector C      0.707
#> 3 Vector A Vector C      0    
foundry_similarity(embeddings, top_k = 1)
#> # A tibble: 1 × 3
#>   text_1   text_2   similarity
#>   <chr>    <chr>         <dbl>
#> 1 Vector A Vector B      0.707
foundry_similarity(embeddings, as_matrix = TRUE)
#>           Vector A  Vector B  Vector C
#> Vector A 1.0000000 0.7071068 0.0000000
#> Vector B 0.7071068 1.0000000 0.7071068
#> Vector C 0.0000000 0.7071068 1.0000000