初始化检索器

hybrid_retriever = HybridRetriever(
embedding_model=bge_model,
weight_vector=0.6, # 向量检索权重
weight_keyword=0.4 # 关键词检索权重
)

添加文档

documents = [
{“doc_id”: “doc1”, “content”: “文档内容1”},
{“doc_id”: “doc2”, “content”: “文档内容2”}
]
hybrid_retriever.add_documents(documents)

搜索

results = hybrid_retriever.search(“查询文本”, top_k=10)

使用熵权法动态调整权重

results = hybrid_retriever.search(
“查询文本”,
top_k=10,
use_entropy_weight=True
)