Struct TfIdfSearch 

pub struct TfIdfSearch { /* private fields */ }

Semantic search without AI (using TF-IDF like approach).

TF-IDF 语义搜索

TF-IDF（Term Frequency-Inverse Document Frequency）是一种 不需要 AI 模型的语义搜索方法。

原理

1. TF（词频）: 词在文档中出现的频率 TF(word, doc) = count(word in doc) / len(doc)

2. IDF（逆文档频率）: 词在整个文档集合中的稀有程度 IDF(word) = log(total_docs / docs_containing_word)

3. TF-IDF: TF × IDF 高 TF-IDF = 词在此文档中重要，但在其他文档中不常见

示例

文档1: "使用 PostgreSQL 数据库"
文档2: "Redis 缓存配置"
文档3: "数据库连接池设置"
 
查询: "数据库"
 
TF-IDF("数据库", 文档1) = 1/3 × log(3/2) = 0.33 × 0.41 = 0.14
TF-IDF("数据库", 文档3) = 1/4 × log(3/2) = 0.25 × 0.41 = 0.10
 
结果: 文档1 > 文档3 > 文档2

Implementations

impl TfIdfSearch

pub fn new() -> Self

Create a new TF-IDF search instance.

pub fn index(&mut self, memory: &AutoMemory)

Index all memories for TF-IDF search.

pub fn search(&self, query: &str, limit: Option<usize>) -> Vec<(String, f32)>

Search using TF-IDF similarity.

pub fn clear(&mut self)

Clear all indices.

Trait Implementations

impl Default for TfIdfSearch

fn default() -> Self

Returns the “default value” for a type.