|
|
|
@ -2,19 +2,26 @@ from WxMini.Milvus.Config.MulvusConfig import *
|
|
|
|
|
from WxMini.Milvus.Utils.MilvusCollectionManager import MilvusCollectionManager
|
|
|
|
|
from WxMini.Milvus.Utils.MilvusConnectionPool import *
|
|
|
|
|
from gensim.models import KeyedVectors
|
|
|
|
|
import jieba # 导入 jieba 分词库
|
|
|
|
|
|
|
|
|
|
# 加载预训练的 Word2Vec 模型
|
|
|
|
|
model_path = "D:/Tencent_AILab_ChineseEmbedding/Tencent_AILab_ChineseEmbedding.txt" # 替换为你的 Word2Vec 模型路径
|
|
|
|
|
model = KeyedVectors.load_word2vec_format(model_path, binary=False, limit=10000)
|
|
|
|
|
print(f"模型加载成功,词向量维度: {model.vector_size}")
|
|
|
|
|
|
|
|
|
|
# 将文本转换为嵌入向量
|
|
|
|
|
def text_to_embedding(text):
|
|
|
|
|
words = text.split()
|
|
|
|
|
words = jieba.lcut(text) # 使用 jieba 分词
|
|
|
|
|
print(f"文本: {text}, 分词结果: {words}")
|
|
|
|
|
embeddings = [model[word] for word in words if word in model]
|
|
|
|
|
print(f"有效词向量数量: {len(embeddings)}")
|
|
|
|
|
if embeddings:
|
|
|
|
|
return sum(embeddings) / len(embeddings) # 取词向量的平均值
|
|
|
|
|
avg_embedding = sum(embeddings) / len(embeddings)
|
|
|
|
|
print(f"生成的平均向量: {avg_embedding[:5]}...") # 打印前 5 维
|
|
|
|
|
return avg_embedding
|
|
|
|
|
else:
|
|
|
|
|
return [0.0] * model.vector_size # 如果文本中没有有效词,返回零向量
|
|
|
|
|
print("未找到有效词,返回零向量")
|
|
|
|
|
return [0.0] * model.vector_size
|
|
|
|
|
|
|
|
|
|
# 1. 使用连接池管理 Milvus 连接
|
|
|
|
|
milvus_pool = MilvusConnectionPool(host=MS_HOST, port=MS_PORT, max_connections=MS_MAX_CONNECTIONS)
|
|
|
|
@ -35,6 +42,10 @@ texts = [
|
|
|
|
|
]
|
|
|
|
|
embeddings = [text_to_embedding(text) for text in texts] # 使用文本模型生成向量
|
|
|
|
|
|
|
|
|
|
# 打印生成的向量值
|
|
|
|
|
for text, embedding in zip(texts, embeddings):
|
|
|
|
|
print(f"文本: {text}, 向量: {embedding[:5]}...") # 打印前 5 维
|
|
|
|
|
|
|
|
|
|
# 5. 插入数据,确保字段顺序与集合定义一致
|
|
|
|
|
entities = [texts, embeddings] # 第一个列表是 text 字段,第二个列表是 embedding 字段
|
|
|
|
|
collection_manager.insert_data(entities)
|
|
|
|
|