Module embeddings

Expand description

向量嵌入模块：提供文本嵌入生成和向量相似度计算功能。

§Embeddings Module

This module provides comprehensive embedding support for AI models, enabling semantic similarity comparisons and vector-based operations.

§Overview

Embeddings are numerical vector representations of text that capture semantic meaning, enabling:

Semantic search and similarity matching
Document clustering and classification
Recommendation systems
Duplicate detection

§Key Features

Feature	Description
Embedding generation	Generate embeddings via AI providers
Vector similarity	Cosine, Euclidean, Manhattan, Dot product metrics
Vector operations	Normalize, average, add, subtract, scale vectors
Similarity search	Find most similar vectors in a collection

§Components

Component	Description
`EmbeddingClient`	Client for generating embeddings from AI providers
`cosine_similarity`	Cosine similarity between vectors (-1 to 1)
`euclidean_distance`	Euclidean (L2) distance between vectors
`find_most_similar`	Find top-k most similar vectors
`normalize_vector`	Normalize vector to unit length

§Example

use ai_lib_rust::embeddings::{cosine_similarity, find_most_similar, SimilarityMetric};

// Calculate similarity between two vectors
let vec_a: Vec<f32> = vec![0.1, 0.2, 0.3];
let vec_b: Vec<f32> = vec![0.15, 0.25, 0.35];
let similarity = cosine_similarity(&vec_a, &vec_b).unwrap();
println!("Similarity: {:.4}", similarity);

// Find most similar vectors in a collection
let query: Vec<f32> = vec![0.5, 0.5, 0.0];
let candidates: Vec<Vec<f32>> = vec![
    vec![0.5, 0.5, 0.0],
    vec![0.0, 0.0, 1.0],
];
let results = find_most_similar(&query, &candidates, 1, SimilarityMetric::Cosine).unwrap();

§Metrics Comparison

Metric	Range	Best For
Cosine	-1 to 1	Semantic similarity (direction)
Euclidean	0 to ∞	Absolute positioning
Dot Product	-∞ to ∞	Magnitude-sensitive comparison
Manhattan	0 to ∞	Sparse vectors, grid distances