Functions — github.com/Chapter-7/Recommended-News

CodeHub MCPcopy Create free account

hub / github.com/Chapter-7/Recommended-News / functions

Functions67 in github.com/Chapter-7/Recommended-News

⨍ Functions 67 ◇ Types & classes 0

↓ 7 callersFunctionget_hist_and_last_click

↓ 7 callersFunctionget_user_item_time

↓ 7 callersFunctionrelease_memory

↓ 7 callersFunctionsubmit

(recall_df, topk=5, model_name=None)

↓ 6 callersFunctionget_item_topk_click

↓ 6 callersFunctionmetrics_recall

(user_recall_items_dict, trn_last_click_df, topk=5)

↓ 6 callersFunctionreduce_mem

↓ 4 callersFunctionitem_based_recommend

基于文章协同过滤的召回 :param user_id: 用户id :param user_item_time_dict: 字典, 根据点击时间获取用户的点击文章序列 {user1: [(item1, time1), (item2, time

↓ 4 callersFunctionnorm_sim

(sim_df, weight=0.0)

↓ 3 callersFunctioncreate_feature

基于用户的历史行为做相关特征 :param users_id: 用户id :param recall_list: 对于每个用户召回的候选文章列表 :param click_hist_df: 用户的历史点击信息 :param articles_inf

↓ 3 callersFunctionget_rank_label_df

(recall_list_df, label_df, is_test=False)

↓ 2 callersFunctiongen_model_input

(train_set, user_profile, seq_max_len)

↓ 2 callersFunctionget_kfold_users

↓ 2 callersFunctionget_user_item_time

↓ 2 callersFunctionneg_sample_recall_data

(recall_items_df, sample_rate=0.001)

↓ 2 callersFunctiontime_diff_mean

↓ 2 callersFunctionuser_based_recommend

基于文章协同过滤的召回 :param user_id: 用户id :param user_item_time_dict: 字典, 根据点击时间获取用户的点击文章序列 {user1: [(item1, time1), (item2, time

↓ 2 callersFunctionyoutubednn_u2i_dict

(data, topk=20)

↓ 1 callersFunctionactive_level

制作区分用户活跃度的特征 :param all_data: 数据集 :param cols: 用到的特征列

↓ 1 callersFunctioncold_start_items

冷启动的情况下召回一些文章 :param user_recall_items_dict: 基于内容embedding相似性召回来的很多文章，字典， {user1: [item1, item2, ..], } :param user_hist_

↓ 1 callersFunctioncombine_recall_results

(user_multi_recall_dict, weight_dict=None, topk=25)

↓ 1 callersFunctiondevice_fea

制作用户的设备特征 :param all_data: 数据集 :param cols: 用到的特征列

↓ 1 callersFunctiongen_data_set

(data, negsample=0)

↓ 1 callersFunctionget_all_click_df

(data_path='../data_raw/', offline=True, sample_size=100000)

↓ 1 callersFunctionget_all_click_df

(data_path='../data/', offline=True)

↓ 1 callersFunctionget_article_info_df

↓ 1 callersFunctionget_click_article_ids_set

↓ 1 callersFunctionget_embedding

(save_path, all_click_df)

↓ 1 callersFunctionget_ensumble_predict_topk

(rank_model, topk=5)

↓ 1 callersFunctionget_hist_and_last_click

↓ 1 callersFunctionget_item_emb_dict

↓ 1 callersFunctionget_item_info_df

↓ 1 callersFunctionget_item_info_dict

↓ 1 callersFunctionget_item_sim_list

↓ 1 callersFunctionget_item_topk_click

↓ 1 callersFunctionget_item_user_time_dict

↓ 1 callersFunctionget_recall_list

(save_path, single_recall_model=None, multi_recall=False)

↓ 1 callersFunctionget_trn_val_tst_data

(data_path, offline=True)

↓ 1 callersFunctionget_user_activate_degree_dict

↓ 1 callersFunctionget_user_hist_item_info_dict

↓ 1 callersFunctionget_user_recall_item_label_df

(click_trn_hist, click_val_hist, click_tst_hist, click_trn_last, click_val_last,

↓ 1 callersFunctionhot_level

制作衡量文章热度的特征 :param all_data: 数据集 :param cols: 用到的特征列

↓ 1 callersFunctionitem_based_recommend

基于文章协同过滤的召回 :param user_id: 用户id :param user_item_time_dict: 字典, 根据点击时间获取用户的点击文章序列[(item1: time1), (item2: time2)..]

↓ 1 callersFunctionitemcf_sim

文章与文章之间的相似性矩阵计算 :param df: 数据表 :item_created_time_dict: 文章创建时间的字典 return : 文章与文章的相似性矩阵思路: 基于物品的协同过滤(详细

↓ 1 callersFunctionmake_item_time_pair

↓ 1 callersFunctionmake_item_time_pair

↓ 1 callersFunctionmake_user_time_pair

↓ 1 callersFunctionmean_diff_time_func

↓ 1 callersFunctionnorm_user_recall_items_sim

(sorted_item_list)

↓ 1 callersFunctionplot_envs

(df, cols, r, c)

↓ 1 callersFunctionrecall_dict_2_df

(recall_list_dict)

↓ 1 callersFunctionsubmit

(recall_df, topk=5, model_name=None)

↓ 1 callersFunctiontrian_item_word2vec

(click_df, embed_size=64, save_name='item_w2v_emb.pkl', split_char=' ')

↓ 1 callersFunctiontrn_val_split

(all_click_df, sample_user_nums)

↓ 1 callersFunctionu2u_embdding_sim

(click_df, user_emb_dict, save_path, topk)

↓ 1 callersFunctionuser_cat_hob_fea

用户的主题爱好 :param all_data: 数据集 :param cols: 用到的特征列

↓ 1 callersFunctionuser_time_hob_fea

制作用户的时间习惯特征 :param all_data: 数据集 :param cols: 用到的特征列

↓ 1 callersFunctionusercf_sim

用户相似性矩阵计算 :param all_click_df: 数据表 :param user_activate_degree_dict: 用户活跃度的字典 return 用户相似性矩阵思路: 基于用户的

Functionembdding_sim

基于内容的文章embedding相似性矩阵计算 :param click_df: 数据表 :param item_emb_df: 文章的embedding :param save_path: 保存路径 :pa

Functionget_all_click_sample

训练集中采样一部分数据调试 data_path: 原数据的存储路径 sample_nums: 采样数目（这里由于机器的内存限制，可以采样用户做）

Functionget_all_click_sample

训练集中采样一部分数据调试 data_path: 原数据的存储路径 sample_nums: 采样数目（这里由于机器的内存限制，可以采样用户做）

Functionhist_func

Functionhist_func

Functionitemcf_sim

文章与文章之间的相似性矩阵计算 :param df: 数据表 :item_created_time_dict: 文章创建时间的字典 return : 文章与文章的相似性矩阵思路: 基于物品的协同过滤(

Functionmake_tuple_func

Functionneg_sample_func

Functionreduce_mem