计算机毕业设计python+spark知识图谱视频推荐系统 视频分析可视化大屏 视频爬虫 视频爬虫 大数据 人工智能 机器学习 深度学习 大数据毕业设计

核心算法代码分享如下：

# -*- codeing = utf-8 -*-
# 创建图谱专用的json文件
#
import numpy as np
import pandas as pd
import json
from db import db_util

d = db_util()
db, cursor = d.get_conn()

def build():
    s_dict = {}
    t_dict = {}
    ret = []

    ind1 = 10000
    ind2 = 20000
    ind3 = 30000
    rind = 900000

    sql = 'select  *  from  tb_detail'
    df = pd.read_sql(sql, con=db)
    for index, row in df.iterrows():

        up_name = row['up_name']
        tags = row['tags'].split(',')[0]

        if up_name not in s_dict:
            ind2 = ind2 + 1
            s_dict[up_name] = ind2
            index2 = ind2
        else:
            index2 = s_dict[up_name]

        if tags not in t_dict:
            ind3 = ind3 + 1
            t_dict[tags] = ind3
            index3 = ind3
        else:
            index3 = t_dict[tags]

        properties = {"name": row['title'], 'coins': row['coins'],  'likes':row['likes'], 'shares':row['shares']}
        start = {'identity': index, 'labels':['视频'], 'properties':properties}
        end = {'identity': index2, 'labels':['UP'], 'properties':{"name": up_name}}
        relationship = {"identity": rind, "start": index,"end": index2,
                        "type": "type", "properties": {"name": "创作"}}
        rind = rind + 1
        segments = []
        segments.append(dict(start=start, relationship=relationship, end=end))

        end = {'identity': index3, 'labels': ['分类'], 'properties': {"name": tags}}
        relationship = {"identity": rind, "start": index, "end": index3,
                        "type": "type", "properties": {"name": "分类"}}
        rind = rind + 1
        segments.append(dict(start=start, relationship=relationship, end=end))

        p = dict(segments=segments, length=1.0)
        ret.append(dict(p=p, score=2))

    json_str = json.dumps(ret, ensure_ascii=False)
    with open('test.json', 'w', encoding='utf8') as f2:
        # ensure_ascii=False才能输入中文，否则是Unicode字符
        # indent=2 JSON数据的缩进，美观
        json.dump(ret, f2, ensure_ascii=False, indent=2)
    print(json_str)
    print("end..")

if __name__ == '__main__':
    build()