项目简介

本项目使用了开源模型，通过 FastApi 提供了一个地址标准化接口

原理

使用了模型MGeo门址地址结构化要素解析-中文-地址领域-base,然后本地化部署了fastapi 服务

from fastapi import FastAPI
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks

from pydantic import BaseModel
import pandas as pd


class Item(BaseModel):
    address_row: str


task = Tasks.token_classification
model = "iic/mgeo_geographic_elements_tagging_chinese_base"

# Initialize the pipeline with the specified task and model
pipeline_ins = pipeline(task=task, model=model, model_revision="master")

app = FastAPI()


# Example endpoint
@app.post("/api/address_standardization")
async def hello(item: Item):
    inputs = item.address_row
    res = pipeline_ins(input=inputs)
    print(res)
    df = pd.DataFrame(res.get("output", []))
    print(df)
    # For now, we just return a placeholder response
    return { row['type']: row['span'] for index, row in df.iterrows()}
  
  
if __name__ == "__main__":
    # test run model
    print("Testing model run 浙江省杭州市余杭区五常街道文一西路969号淘宝城5号楼，放前台")
    res = pipeline_ins(input="浙江省杭州市余杭区五常街道文一西路969号淘宝城5号楼，放前台")
    print(res)

usage 使用方法

构建镜像

docker build . -t address_model

启动

docker compose up -d

测试