推荐系统算法示例

# word2vec
# https://github.com/michaelldd/recommend_system_learning/blob/master/word2vecor_recommend_main.py

# KNN
# https://github.com/michaelldd/recommend_system_learning/blob/master/surprise_recommend_main.py

# 物品协同过滤
# https://github.com/LuckyZXL2016/Movie_Recommend/blob/master/Spark_Movie/src/main/scala/com/zxl/ModelTrain/ItemSimilarity.scala
# 同现，余弦，欧几里德

# 矩阵分解: ALS, SVD
# https://github.com/microsoft/recommenders/blob/main/examples/02_model_collaborative_filtering/surprise_svd_deep_dive.ipynb
# ALS: https://blog.csdn.net/weixin_30533797/article/details/98616142

# 内容过滤：lightGbm

# 混合过滤：lightfm, libfm, libffm, xlearn
# https://github.com/microsoft/recommenders/blob/efaa3d7742183dee0846877e2dc64977098e1977/examples/02_model_hybrid/fm_deep_dive.ipynb


# NLP
# bert
# https://github.com/NLP-LOVE/ML-NLP/blob/master/NLP/16.8%20BERT/bert-Chinese-classification-task.md

# XLNet
# https://github.com/ymcui/Chinese-XLNet

# ItemBaseCf
# https://github.com/HanXiaoyang/pyspark-recommendation-demo/blob/2d8521dad38b5988a12ed3a1fdf4a3b135d6d371/itemBasedRecommender.py#L84

# UserBaseCf
# https://github.com/HanXiaoyang/pyspark-recommendation-demo/blob/master/userBasedRecommender.py

# XGBoost调优
# https://eason.blog.csdn.net/article/details/87313329

# Xgboost, LR ctr
# https://github.com/blogchong/data_and_advertisement/blob/master/code/017_ctr_xgboost_lr/03_feature_xgboost.ipynb

# 基于Spark的GBDT + LR模型实现
# https://www.icode9.com/content-4-113052.html

# spark-xgboost 8.1
# https://blog.csdn.net/weixin_39562615/article/details/111674845

# lsh
# https://github.com/soundcloud/cosine-lsh-join-spark

# 余弦相似公式
# def consineSimilarity(movie1: DoubleMatrix, movie2: DoubleMatrix): Double = {
#     movie1.dot(movie2) / (movie1.norm2() * movie2.norm2())
# }


# hdfs api:
# def getVideoTagsM(inputDir: String) =
# {
#     val
# videoTagsM = mutable.HashMap[String, mutable.HashSet[String]]()
# val
# hdfs: FileSystem = FileSystem.get(new
# Configuration())
# val
# path = new
# Path(inputDir)
#
# if (hdfs.isDirectory(path))
# {
# for (status < - hdfs.listStatus(new Path(inputDir))) {
# val fpath = status.getPath
# var bufferedReader = new BufferedReader(new InputStreamReader(hdfs.open(fpath)))
# getVideoTagsMFromF(bufferedReader, videoTagsM)
# }
# } else {
# var bufferedReader = new BufferedReader(new InputStreamReader(hdfs.open(path)))
# getVideoTagsMFromF(bufferedReader, videoTagsM)
# }
# videoTagsM
# }


# write hdfs api
# try {
# val outputObject = new Path( outputPath + "/object")
# val hdfs: FileSystem = FileSystem.get(new
# Configuration)
# val outputstream: FSDataOutputStream = hdfs.create(outputObject)
# val
# oos = new
# ObjectOutputStream(outputstream)
# oos.writeObject(model);
# oos.close();
# } catch
# {
#     case
# e: Exception = > println("Self Expception:" + e.toString())
# }