title_with_video/applications/match_algorithm/title_similarity.py

"""
@author: luojunhui
"""
import aiohttp

empty_list = []


def jcd_title_similarity(ori_title, search_title):
    """
    simple ways to calculate the similarity of titles
    :param ori_title:
    :param search_title:
    :return:
    """
    set1 = set(ori_title)
    set2 = set(search_title)
    intersection = len(set1 & set2)
    union = len(set1 | set2)
    return intersection / union


async def nlp_title_similarity(url, ori_title, search_title_list):
    """
    nlp title similarity
    """
    headers = {"Content-Type": "application/json"}
    body = {
        "data": {
            "text_list_a": [ori_title],
            "text_list_b": search_title_list,
        },
        "function": "similarities_cross",
        "use_cache": False
    }
    async with aiohttp.ClientSession() as session:
        async with session.post(url, headers=headers, json=body) as response:
            response_text = await response.text()
            if response_text and response.status == 200:
                res = await response.json()
                score_list = res['score_list_list'][0]
                return score_list
            else:
                return empty_list