LongArticleAlgServer/applications/textSimilarity.py

"""
@author: luojunhui
"""
import time
import torch
import numpy as np
from similarities import BertSimilarity


# bge_large_zh_v1_5 = 'bge_large_zh_v1_5'
# text2vec_base_chinese = "text2vec_base_chinese"
# text2vec_bge_large_chinese = "text2vec_bge_large_chinese"


class NLPFunction(object):
    """
    NLP Task
    """

    def __init__(self, model):
        self.model = model

    def base_string_similarity(self, text_dict):
        """
        基础功能，计算两个字符串的相似度
        :param text_dict:
        :return:
        """
        score_tensor = self.model.similarity(
            text_dict['text_a'],
            text_dict['text_b']
        )
        return score_tensor.squeeze().tolist()

    def base_list_similarity(self, pair_list_dict):
        """
        计算两个list的相似度
        :return: "score_list_b": [100, 1000, 500, 40],
        """
        score_tensor = self.model.similarity(
            pair_list_dict['text_list_a'],
            pair_list_dict['text_list_b']
        )
        return score_tensor.tolist()


if __name__ == '__main__':
    a = time.time()
    m = BertSimilarity(model_name_or_path="BAAI/bge-large-zh-v1.5")
    b = time.time()
    print("模型加载时间：\t", b - a)
    NF = NLPFunction(m)
    td = {
        "text_a": "王者荣耀",
        "text_b": "斗罗大陆"
    }
    tld = {
        "text_list_a": ["凯旋", "圣洁", "篮球"],
        "text_list_b": ["胜利", "纯洁", "足球"]
    }
    # res = NF.base_string_similarity(text_dict=td)
    res = NF.base_list_similarity(pair_list_dict=tld)
    c = time.time()
    print("计算时间：\t", c - b)
    for i in res:
        print(i)