milvus入门使用

插入数据后的效果：

 代码如下：

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86

import configparser from pymilvus import connections, Collection, DataType, FieldSchema, CollectionSchema import numpy as np   def create_collection():     # Define the schema     fields = [         FieldSchema(name="sentence_id", dtype=DataType.INT64, is_primary=True, auto_id=True),         FieldSchema(name="sentence", dtype=DataType.VARCHAR, max_length=512),         FieldSchema(name="embedding", dtype=DataType.FLOAT_VECTOR, dim=128)     ]     schema = CollectionSchema(fields, description="Sentence collection")       # Create the collection     collection = Collection(name="sentence_collection", schema=schema)     return collection   def insert_data(collection):     sentences = [         "这是第一句。",         "这是第二句。",         "这是第三句。"     ]           embeddings = np.random.rand(len(sentences), 128).tolist()  # Generate 128-dimensional vectors           entities = [         sentences,         embeddings     ]       insert_result = collection.insert(entities)     print(f"Inserted {len(insert_result.primary_keys)} records into collection.")   def create_index(collection):     index_params = {         "index_type": "IVF_FLAT",         "params": {"nlist": 128},         "metric_type": "L2"     }     collection.create_index(field_name="embedding", index_params=index_params)     print("Index created.")   def search_data(collection, query_sentence):     query_embedding = np.random.rand(1, 128).tolist()  # Generate a vector for the query sentence       search_params = {"metric_type": "L2", "params": {"nprobe": 10}}           results = collection.search(         data=query_embedding,         anns_field="embedding",         param=search_params,         limit=3,         expr=None,         output_fields=["sentence"]     )           for hits in results:         for hit in hits:             print(f"Match found: {hit.id} with distance: {hit.distance}, sentence: {hit.entity.get('sentence')}")   if __name__ == '__main__':     # Connect to Milvus     cfp = configparser.RawConfigParser()     cfp.read('config.ini')     milvus_uri = cfp.get('example', 'uri')     token = cfp.get('example', 'token')     connections.connect("default",                         uri=milvus_uri,                         token=token)     print(f"Connecting to DB: {milvus_uri}")           # Create collection     collection = create_collection()       # Insert data     insert_data(collection)           # Create index     create_index(collection)           # Load the collection into memory     collection.load()           # Search data     search_data(collection, "这是一个查询句子。")

　　

运行效果：

python hello_zilliz_vectordb.py
Connecting to DB: https://in03-ca69f49bb65709f.api.gcp-us-west1.zillizcloud.com
Inserted 3 records into collection.
Index created.
Match found: 450140263656791260 with distance: 19.557846069335938, sentence: 这是第二句。
Match found: 450140263656791261 with distance: 20.327802658081055, sentence: 这是第三句。
Match found: 450140263656791259 with distance: 20.40052032470703, sentence: 这是第一句。

 

注意事项：

1. 向量转换：上面的代码使用了随机向量来模拟句子向量。在实际应用中，您需要使用 NLP 模型（例如中文 BERT）来将中文句子转换为向量。
2. 字符编码：确保在读取和处理中文文本时使用正确的字符编码（通常是 UTF-8）。