Performs semantic analysis including similarity, dimensionality reduction, and clustering in a single call.
Usage
fit_semantic_model(
texts,
analysis_types = c("similarity", "dimensionality_reduction", "clustering"),
document_feature_type = "embeddings",
similarity_method = "cosine",
use_embeddings = TRUE,
embedding_model = "all-MiniLM-L6-v2",
dimred_method = "UMAP",
clustering_method = "umap_dbscan",
n_components = 2,
n_clusters = 5,
seed = 123,
verbose = TRUE
)Arguments
- texts
A character vector of texts to analyze.
- analysis_types
Types of analysis to perform: "similarity", "dimensionality_reduction", "clustering".
- document_feature_type
Feature extraction type (default: "embeddings").
- similarity_method
Similarity calculation method (default: "cosine").
- use_embeddings
Logical, use embedding-based approaches (default: TRUE).
- embedding_model
Sentence transformer model name (default: "all-MiniLM-L6-v2").
- dimred_method
Dimensionality reduction method: "PCA", "t-SNE", "UMAP" (default: "UMAP").
- clustering_method
Clustering method: "kmeans", "hierarchical", "umap_dbscan" (default: "umap_dbscan").
- n_components
Number of dimensions for reduction (default: 2).
- n_clusters
Number of clusters (default: 5).
- seed
Random seed for reproducibility (default: 123).
- verbose
Logical, if TRUE, prints progress messages.
Examples
if (interactive()) {
data(SpecialEduTech)
texts <- SpecialEduTech$abstract[1:10]
results <- fit_semantic_model(
texts = texts,
analysis_types = c("similarity", "clustering")
)
print(results)
}
#> Starting semantic analysis...
#> Analysis types: similarity, clustering
#> Step 1: Document similarity analysis...
#> Starting document similarity analysis...
#> Feature type: embeddings
#> Similarity method: cosine
#> Use embeddings: TRUE
#> Step 1: Generating feature matrix...
#> Using embedding features...
#> Processing 10 documents with embeddings...
#> Step 2: Calculating similarity matrix...
#> Step 3: Calculating metrics...
#> Document similarity analysis completed in 2.48 seconds
#> Documents analyzed: 10
#> Feature dimensions: 384
#> Step 3: Clustering analysis...
#> Starting clustering analysis with method: umap_dbscan
#> Performing UMAP + DBSCAN clustering...
#> Starting dimensionality reduction with method: UMAP
#> Performing UMAP on original features...
#> Warning: failed creating initial embedding; using random embedding instead
#> Dimensionality reduction completed in 0.26 seconds
#> Determining optimal eps parameter...
#> Error in value[[3L]](cond): Error in clustering analysis: Not enough neighbors in data set!