基于用户的协同过滤算法改进（usercf 改进）

usercf 协同过滤算法改进

 2018/11/22   Share

调试流程

def main_flow():
    """
    main flow of itmecf
    :return:
    """
    user_click, user_click_time = reader.get_user_click(
        "/home/pzs741/PycharmProjects/CollaborativeFiltering/data/ratings.csv")
    item_info = reader.get_item_info(
        "/home/pzs741/PycharmProjects/CollaborativeFiltering/data/movies.csv")
    item_click_by_user = transfer_user_click(user_click)
    user_sim = cal_user_sim(item_click_by_user,user_click_time)
    debug_user_sim(user_sim)
    # recom_result = cal_recom_result(user_click,user_sim)
    # debug_recom_result(item_info,recom_result)

调试代码

用户相似度调试代码


def debug_user_sim(user_sim):
    """
    print user sim result
    :param user_sim:key userid value:[(userid1,score1),(userid2,score2)] 
    :return: 
    """
    topk = 5
    fix_user = "1"
    if fix_user not in user_sim:
        print("invalid user")
        return
    for zuhe in user_sim[fix_user][:topk]:
        userid, score = zuhe
        print(fix_user + "\tsim_user" + userid + "\t" + str(score))

用户推荐结果调试代码

def debug_recom_result(item_info, recom_result):
    """
    print recom result for user
    :param item_info: key itemid value:[title,genres]
    :param recom_result: key userid value dict, value_key:itemid value_value:recom_result
    :return:
    """
    fix_user = "1"
    if fix_user not in recom_result:
        print("invalid user for recoming result")
        return
    for itemid in recom_result["1"]:
        if itemid not in item_info:
            continue
        recom_score = recom_result["1"][itemid]
        print("recom_result:" + ",".join(item_info[itemid]) + "\t" + str(recom_score))

算法改进1

def update_contribution_score(item_user_click_count):
    """
    usercf user contribution score update v1
    :param item_user_click_count: how many user have click this item
    :return: contribution score
    """
    return 1 / math.log10(1 + item_user_click_count)

算法改进2

def update_two_contribution_score(click_time_one, click_time_two):
    """
    user cf user contribution score update v2
    :param click_time_one:
    :param click_time_two:
    :return: contribution score
    """
    delta_time = abs(click_time_two-click_time_one)
    total_sec = 60 * 60 * 24
    delta_time = delta_time / total_sec
    return 1/(1 + delta_time)

调试结果

改进1的调试结果

用户相似度展示
1 sim_user313 0.18803307299446
1 sim_user57 0.1879004753996153
1 sim_user288 0.18756012667350208
1 sim_user266 0.18399642220363496
1 sim_user368 0.17954277437174224

推荐结果展示
recom_result:Heat (1995),Action|Crime|Thriller 0.18803307299446
recom_result:GoldenEye (1995),Action|Adventure|Thriller 0.18803307299446
recom_result:“City of Lost Children, The (Cité des enfants perdus, La) (1995)”,Adventure|Drama|Fantasy|Mystery|Sci-Fi 0.18803307299446
recom_result:Twelve Monkeys (a.k.a. 12 Monkeys) (1995),Mystery|Sci-Fi|Thriller 0.18803307299446
recom_result:Clueless (1995),Comedy|Romance 0.18803307299446
recom_result:Toy Story (1995),Adventure|Animation|Children|Comedy|Fantasy 0.1879004753996153
recom_result:“American President, The (1995)”,Comedy|Drama|Romance 0.1879004753996153
recom_result:Get Shorty (1995),Comedy|Crime|Thriller 0.1879004753996153
recom_result:Grumpier Old Men (1995),Comedy|Romance 0.18756012667350208
recom_result:Sense and Sensibility (1995),Drama|Romance 0.18756012667350208

改进2的调试结果

用户相似度展示
1 sim_user96 0.15791081894849407
1 sim_user469 0.1172771823445635
1 sim_user27 0.08282266726133267
1 sim_user265 0.07282310151619956
1 sim_user19 0.06379553900727523

推荐结果展示
recom_result:Toy Story (1995),Adventure|Animation|Children|Comedy|Fantasy 0.15791081894849407
recom_result:Babe (1995),Children|Drama 0.15791081894849407
recom_result:“Usual Suspects, The (1995)”,Crime|Mystery|Thriller 0.15791081894849407
recom_result:Braveheart (1995),Action|Drama|War 0.15791081894849407
recom_result:Apollo 13 (1995),Adventure|Drama|IMAX 0.15791081894849407
recom_result:Heat (1995),Action|Crime|Thriller 0.1172771823445635
recom_result:“American President, The (1995)”,Comedy|Drama|Romance 0.1172771823445635
recom_result:“City of Lost Children, The (Cité des enfants perdus, La) (1995)”,Adventure|Drama|Fantasy|Mystery|Sci-Fi 0.1172771823445635
recom_result:Twelve Monkeys (a.k.a. 12 Monkeys) (1995),Mystery|Sci-Fi|Thriller 0.1172771823445635
recom_result:Jumanji (1995),Adventure|Children|Fantasy 0.08282266726133267
recom_result:Pocahontas (1995),Animation|Children|Drama|Musical|Romance 0.08282266726133267
recom_result:“Indian in the Cupboard, The (1995)”,Adventure|Children|Fantasy 0.08282266726133267

Next Post

TensorFlow与Flask结合打造手写体数字识别
Previous Post

基于物品的协同过滤算法改进（itemcf 改进）

CATALOG

1. 调试流程
2. 调试代码
3. 算法改进1
4. 算法改进2
5. 调试结果



缺失模块。
1、请确保node版本大于6.2
2、在博客根目录（注意不是archer根目录）执行以下命令：
npm i hexo-generator-json-content --save
3、在根目录_config.yml里添加配置：

jsonContent:
  meta: false
  pages: false
  posts:
    title: true
    date: true
    path: true
    text: false
    raw: false
    content: false
    slug: false
    updated: false
    comments: false
    link: false
    permalink: false
    excerpt: false
    categories: true
    tags: true