You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hivemall.apache.org by my...@apache.org on 2017/09/28 03:28:05 UTC
[07/14] incubator-hivemall-site git commit: Updated userguide for SLIM
http://git-wip-us.apache.org/repos/asf/incubator-hivemall-site/blob/4c02d418/userguide/recommend/movielens_slim.html
----------------------------------------------------------------------
diff --git a/userguide/recommend/movielens_slim.html b/userguide/recommend/movielens_slim.html
new file mode 100644
index 0000000..743545e
--- /dev/null
+++ b/userguide/recommend/movielens_slim.html
@@ -0,0 +1,2952 @@
+
+<!DOCTYPE HTML>
+<html lang="" >
+ <head>
+ <meta charset="UTF-8">
+ <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
+ <title>SLIM for Fast Top-K Recommendation ยท Hivemall User Manual</title>
+ <meta http-equiv="X-UA-Compatible" content="IE=edge" />
+ <meta name="description" content="">
+ <meta name="generator" content="GitBook 3.2.2">
+
+
+
+
+ <link rel="stylesheet" href="../gitbook/style.css">
+
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-splitter/splitter.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-etoc/plugin.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-callouts/plugin.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-toggle-chapters/toggle.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-codeblock-filename/block.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-expandable-chapters/expandable-chapters.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-multipart/multipart.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-katex/katex.min.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-emphasize/plugin.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-highlight/website.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-search/search.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-fontsettings/website.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-theme-api/theme-api.css">
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ <meta name="HandheldFriendly" content="true"/>
+ <meta name="viewport" content="width=device-width, initial-scale=1, user-scalable=no">
+ <meta name="apple-mobile-web-app-capable" content="yes">
+ <meta name="apple-mobile-web-app-status-bar-style" content="black">
+ <link rel="apple-touch-icon-precomposed" sizes="152x152" href="../gitbook/images/apple-touch-icon-precomposed-152.png">
+ <link rel="shortcut icon" href="../gitbook/images/favicon.ico" type="image/x-icon">
+
+
+ <link rel="next" href="movielens_cv.html" />
+
+
+ <link rel="prev" href="movielens_fm.html" />
+
+
+ </head>
+ <body>
+
+<div class="book">
+ <div class="book-summary">
+
+
+<div id="book-search-input" role="search">
+ <input type="text" placeholder="Type to search" />
+</div>
+
+
+ <nav role="navigation">
+
+
+
+<ul class="summary">
+
+
+
+
+ <li>
+ <a href="http://hivemall.incubator.apache.org/" target="_blank" class="custom-link"><i class="fa fa-home"></i> Home</a>
+ </li>
+
+
+
+
+ <li class="divider"></li>
+
+
+
+
+ <li class="header">TABLE OF CONTENTS</li>
+
+
+
+ <li class="chapter " data-level="1.1" data-path="../">
+
+ <a href="../">
+
+
+ <b>1.1.</b>
+
+ Introduction
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.2" data-path="../getting_started/">
+
+ <a href="../getting_started/">
+
+
+ <b>1.2.</b>
+
+ Getting Started
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.2.1" data-path="../getting_started/installation.html">
+
+ <a href="../getting_started/installation.html">
+
+
+ <b>1.2.1.</b>
+
+ Installation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.2.2" data-path="../getting_started/permanent-functions.html">
+
+ <a href="../getting_started/permanent-functions.html">
+
+
+ <b>1.2.2.</b>
+
+ Install as permanent functions
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.2.3" data-path="../getting_started/input-format.html">
+
+ <a href="../getting_started/input-format.html">
+
+
+ <b>1.2.3.</b>
+
+ Input Format
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="1.3" data-path="../tips/">
+
+ <a href="../tips/">
+
+
+ <b>1.3.</b>
+
+ Tips for Effective Hivemall
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.3.1" data-path="../tips/addbias.html">
+
+ <a href="../tips/addbias.html">
+
+
+ <b>1.3.1.</b>
+
+ Explicit add_bias() for better prediction
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.2" data-path="../tips/rand_amplify.html">
+
+ <a href="../tips/rand_amplify.html">
+
+
+ <b>1.3.2.</b>
+
+ Use rand_amplify() to better prediction results
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.3" data-path="../tips/rt_prediction.html">
+
+ <a href="../tips/rt_prediction.html">
+
+
+ <b>1.3.3.</b>
+
+ Real-time Prediction on RDBMS
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.4" data-path="../tips/ensemble_learning.html">
+
+ <a href="../tips/ensemble_learning.html">
+
+
+ <b>1.3.4.</b>
+
+ Ensemble learning for stable prediction
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.5" data-path="../tips/mixserver.html">
+
+ <a href="../tips/mixserver.html">
+
+
+ <b>1.3.5.</b>
+
+ Mixing models for a better prediction convergence (MIX server)
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.6" data-path="../tips/emr.html">
+
+ <a href="../tips/emr.html">
+
+
+ <b>1.3.6.</b>
+
+ Run Hivemall on Amazon Elastic MapReduce
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="1.4" data-path="../tips/general_tips.html">
+
+ <a href="../tips/general_tips.html">
+
+
+ <b>1.4.</b>
+
+ General Hive/Hadoop tips
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.4.1" data-path="../tips/rowid.html">
+
+ <a href="../tips/rowid.html">
+
+
+ <b>1.4.1.</b>
+
+ Adding rowid for each row
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.4.2" data-path="../tips/hadoop_tuning.html">
+
+ <a href="../tips/hadoop_tuning.html">
+
+
+ <b>1.4.2.</b>
+
+ Hadoop tuning for Hivemall
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="1.5" data-path="../troubleshooting/">
+
+ <a href="../troubleshooting/">
+
+
+ <b>1.5.</b>
+
+ Troubleshooting
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.5.1" data-path="../troubleshooting/oom.html">
+
+ <a href="../troubleshooting/oom.html">
+
+
+ <b>1.5.1.</b>
+
+ OutOfMemoryError in training
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.2" data-path="../troubleshooting/mapjoin_task_error.html">
+
+ <a href="../troubleshooting/mapjoin_task_error.html">
+
+
+ <b>1.5.2.</b>
+
+ SemanticException Generate Map Join Task Error: Cannot serialize object
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.3" data-path="../troubleshooting/asterisk.html">
+
+ <a href="../troubleshooting/asterisk.html">
+
+
+ <b>1.5.3.</b>
+
+ Asterisk argument for UDTF does not work
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.4" data-path="../troubleshooting/num_mappers.html">
+
+ <a href="../troubleshooting/num_mappers.html">
+
+
+ <b>1.5.4.</b>
+
+ The number of mappers is less than input splits in Hadoop 2.x
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.5" data-path="../troubleshooting/mapjoin_classcastex.html">
+
+ <a href="../troubleshooting/mapjoin_classcastex.html">
+
+
+ <b>1.5.5.</b>
+
+ Map-side Join causes ClassCastException on Tez
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part II - Generic Features</li>
+
+
+
+ <li class="chapter " data-level="2.1" data-path="../misc/generic_funcs.html">
+
+ <a href="../misc/generic_funcs.html">
+
+
+ <b>2.1.</b>
+
+ List of generic Hivemall functions
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="2.2" data-path="../misc/topk.html">
+
+ <a href="../misc/topk.html">
+
+
+ <b>2.2.</b>
+
+ Efficient Top-K query processing
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="2.3" data-path="../misc/tokenizer.html">
+
+ <a href="../misc/tokenizer.html">
+
+
+ <b>2.3.</b>
+
+ Text Tokenizer
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part III - Feature Engineering</li>
+
+
+
+ <li class="chapter " data-level="3.1" data-path="../ft_engineering/scaling.html">
+
+ <a href="../ft_engineering/scaling.html">
+
+
+ <b>3.1.</b>
+
+ Feature Scaling
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.2" data-path="../ft_engineering/hashing.html">
+
+ <a href="../ft_engineering/hashing.html">
+
+
+ <b>3.2.</b>
+
+ Feature Hashing
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.3" data-path="../ft_engineering/selection.html">
+
+ <a href="../ft_engineering/selection.html">
+
+
+ <b>3.3.</b>
+
+ Feature Selection
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.4" data-path="../ft_engineering/binning.html">
+
+ <a href="../ft_engineering/binning.html">
+
+
+ <b>3.4.</b>
+
+ Feature Binning
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.5" data-path="../ft_engineering/pairing.html">
+
+ <a href="../ft_engineering/pairing.html">
+
+
+ <b>3.5.</b>
+
+ FEATURE PAIRING
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="3.5.1" data-path="../ft_engineering/polynomial.html">
+
+ <a href="../ft_engineering/polynomial.html">
+
+
+ <b>3.5.1.</b>
+
+ Polynomial Features
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="3.6" data-path="../ft_engineering/ft_trans.html">
+
+ <a href="../ft_engineering/ft_trans.html">
+
+
+ <b>3.6.</b>
+
+ FEATURE TRANSFORMATION
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="3.6.1" data-path="../ft_engineering/vectorization.html">
+
+ <a href="../ft_engineering/vectorization.html">
+
+
+ <b>3.6.1.</b>
+
+ Feature Vectorization
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.6.2" data-path="../ft_engineering/quantify.html">
+
+ <a href="../ft_engineering/quantify.html">
+
+
+ <b>3.6.2.</b>
+
+ Quantify non-number features
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="3.7" data-path="../ft_engineering/tfidf.html">
+
+ <a href="../ft_engineering/tfidf.html">
+
+
+ <b>3.7.</b>
+
+ TF-IDF Calculation
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part IV - Evaluation</li>
+
+
+
+ <li class="chapter " data-level="4.1" data-path="../eval/binary_classification_measures.html">
+
+ <a href="../eval/binary_classification_measures.html">
+
+
+ <b>4.1.</b>
+
+ Binary Classification Metrics
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="4.1.1" data-path="../eval/auc.html">
+
+ <a href="../eval/auc.html">
+
+
+ <b>4.1.1.</b>
+
+ Area Under the ROC Curve
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="4.2" data-path="../eval/multilabel_classification_measures.html">
+
+ <a href="../eval/multilabel_classification_measures.html">
+
+
+ <b>4.2.</b>
+
+ Multi-label Classification Metrics
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="4.3" data-path="../eval/regression.html">
+
+ <a href="../eval/regression.html">
+
+
+ <b>4.3.</b>
+
+ Regression metrics
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="4.4" data-path="../eval/rank.html">
+
+ <a href="../eval/rank.html">
+
+
+ <b>4.4.</b>
+
+ Ranking Measures
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="4.5" data-path="../eval/datagen.html">
+
+ <a href="../eval/datagen.html">
+
+
+ <b>4.5.</b>
+
+ Data Generation
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="4.5.1" data-path="../eval/lr_datagen.html">
+
+ <a href="../eval/lr_datagen.html">
+
+
+ <b>4.5.1.</b>
+
+ Logistic Regression data generation
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part V - Supervised Learning</li>
+
+
+
+ <li class="chapter " data-level="5.1" data-path="../misc/prediction.html">
+
+ <a href="../misc/prediction.html">
+
+
+ <b>5.1.</b>
+
+ How Prediction Works
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part VI - Binary classification</li>
+
+
+
+ <li class="chapter " data-level="6.1" data-path="../binaryclass/general.html">
+
+ <a href="../binaryclass/general.html">
+
+
+ <b>6.1.</b>
+
+ Binary Classification
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.2" data-path="../binaryclass/a9a.html">
+
+ <a href="../binaryclass/a9a.html">
+
+
+ <b>6.2.</b>
+
+ a9a tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.2.1" data-path="../binaryclass/a9a_dataset.html">
+
+ <a href="../binaryclass/a9a_dataset.html">
+
+
+ <b>6.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.2.2" data-path="../binaryclass/a9a_lr.html">
+
+ <a href="../binaryclass/a9a_lr.html">
+
+
+ <b>6.2.2.</b>
+
+ Logistic Regression
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.2.3" data-path="../binaryclass/a9a_minibatch.html">
+
+ <a href="../binaryclass/a9a_minibatch.html">
+
+
+ <b>6.2.3.</b>
+
+ Mini-batch Gradient Descent
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.3" data-path="../binaryclass/news20.html">
+
+ <a href="../binaryclass/news20.html">
+
+
+ <b>6.3.</b>
+
+ News20 tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.3.1" data-path="../binaryclass/news20_dataset.html">
+
+ <a href="../binaryclass/news20_dataset.html">
+
+
+ <b>6.3.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.3.2" data-path="../binaryclass/news20_pa.html">
+
+ <a href="../binaryclass/news20_pa.html">
+
+
+ <b>6.3.2.</b>
+
+ Perceptron, Passive Aggressive
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.3.3" data-path="../binaryclass/news20_scw.html">
+
+ <a href="../binaryclass/news20_scw.html">
+
+
+ <b>6.3.3.</b>
+
+ CW, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.3.4" data-path="../binaryclass/news20_adagrad.html">
+
+ <a href="../binaryclass/news20_adagrad.html">
+
+
+ <b>6.3.4.</b>
+
+ AdaGradRDA, AdaGrad, AdaDelta
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.3.5" data-path="../binaryclass/news20_rf.html">
+
+ <a href="../binaryclass/news20_rf.html">
+
+
+ <b>6.3.5.</b>
+
+ Random Forest
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.4" data-path="../binaryclass/kdd2010a.html">
+
+ <a href="../binaryclass/kdd2010a.html">
+
+
+ <b>6.4.</b>
+
+ KDD2010a tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.4.1" data-path="../binaryclass/kdd2010a_dataset.html">
+
+ <a href="../binaryclass/kdd2010a_dataset.html">
+
+
+ <b>6.4.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.4.2" data-path="../binaryclass/kdd2010a_scw.html">
+
+ <a href="../binaryclass/kdd2010a_scw.html">
+
+
+ <b>6.4.2.</b>
+
+ PA, CW, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.5" data-path="../binaryclass/kdd2010b.html">
+
+ <a href="../binaryclass/kdd2010b.html">
+
+
+ <b>6.5.</b>
+
+ KDD2010b tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.5.1" data-path="../binaryclass/kdd2010b_dataset.html">
+
+ <a href="../binaryclass/kdd2010b_dataset.html">
+
+
+ <b>6.5.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.5.2" data-path="../binaryclass/kdd2010b_arow.html">
+
+ <a href="../binaryclass/kdd2010b_arow.html">
+
+
+ <b>6.5.2.</b>
+
+ AROW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.6" data-path="../binaryclass/webspam.html">
+
+ <a href="../binaryclass/webspam.html">
+
+
+ <b>6.6.</b>
+
+ Webspam tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.6.1" data-path="../binaryclass/webspam_dataset.html">
+
+ <a href="../binaryclass/webspam_dataset.html">
+
+
+ <b>6.6.1.</b>
+
+ Data pareparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.6.2" data-path="../binaryclass/webspam_scw.html">
+
+ <a href="../binaryclass/webspam_scw.html">
+
+
+ <b>6.6.2.</b>
+
+ PA1, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.7" data-path="../binaryclass/titanic_rf.html">
+
+ <a href="../binaryclass/titanic_rf.html">
+
+
+ <b>6.7.</b>
+
+ Kaggle Titanic tutorial
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part VII - Multiclass classification</li>
+
+
+
+ <li class="chapter " data-level="7.1" data-path="../multiclass/news20.html">
+
+ <a href="../multiclass/news20.html">
+
+
+ <b>7.1.</b>
+
+ News20 Multiclass tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="7.1.1" data-path="../multiclass/news20_dataset.html">
+
+ <a href="../multiclass/news20_dataset.html">
+
+
+ <b>7.1.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.2" data-path="../multiclass/news20_one-vs-the-rest_dataset.html">
+
+ <a href="../multiclass/news20_one-vs-the-rest_dataset.html">
+
+
+ <b>7.1.2.</b>
+
+ Data preparation for one-vs-the-rest classifiers
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.3" data-path="../multiclass/news20_pa.html">
+
+ <a href="../multiclass/news20_pa.html">
+
+
+ <b>7.1.3.</b>
+
+ PA
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.4" data-path="../multiclass/news20_scw.html">
+
+ <a href="../multiclass/news20_scw.html">
+
+
+ <b>7.1.4.</b>
+
+ CW, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.5" data-path="../multiclass/news20_ensemble.html">
+
+ <a href="../multiclass/news20_ensemble.html">
+
+
+ <b>7.1.5.</b>
+
+ Ensemble learning
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.6" data-path="../multiclass/news20_one-vs-the-rest.html">
+
+ <a href="../multiclass/news20_one-vs-the-rest.html">
+
+
+ <b>7.1.6.</b>
+
+ one-vs-the-rest classifier
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="7.2" data-path="../multiclass/iris.html">
+
+ <a href="../multiclass/iris.html">
+
+
+ <b>7.2.</b>
+
+ Iris tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="7.2.1" data-path="../multiclass/iris_dataset.html">
+
+ <a href="../multiclass/iris_dataset.html">
+
+
+ <b>7.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.2.2" data-path="../multiclass/iris_scw.html">
+
+ <a href="../multiclass/iris_scw.html">
+
+
+ <b>7.2.2.</b>
+
+ SCW
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.2.3" data-path="../multiclass/iris_randomforest.html">
+
+ <a href="../multiclass/iris_randomforest.html">
+
+
+ <b>7.2.3.</b>
+
+ Random Forest
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part VIII - Regression</li>
+
+
+
+ <li class="chapter " data-level="8.1" data-path="../regression/general.html">
+
+ <a href="../regression/general.html">
+
+
+ <b>8.1.</b>
+
+ Regression
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.2" data-path="../regression/e2006.html">
+
+ <a href="../regression/e2006.html">
+
+
+ <b>8.2.</b>
+
+ E2006-tfidf regression tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="8.2.1" data-path="../regression/e2006_dataset.html">
+
+ <a href="../regression/e2006_dataset.html">
+
+
+ <b>8.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.2.2" data-path="../regression/e2006_arow.html">
+
+ <a href="../regression/e2006_arow.html">
+
+
+ <b>8.2.2.</b>
+
+ Passive Aggressive, AROW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="8.3" data-path="../regression/kddcup12tr2.html">
+
+ <a href="../regression/kddcup12tr2.html">
+
+
+ <b>8.3.</b>
+
+ KDDCup 2012 track 2 CTR prediction tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="8.3.1" data-path="../regression/kddcup12tr2_dataset.html">
+
+ <a href="../regression/kddcup12tr2_dataset.html">
+
+
+ <b>8.3.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.3.2" data-path="../regression/kddcup12tr2_lr.html">
+
+ <a href="../regression/kddcup12tr2_lr.html">
+
+
+ <b>8.3.2.</b>
+
+ Logistic Regression, Passive Aggressive
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.3.3" data-path="../regression/kddcup12tr2_lr_amplify.html">
+
+ <a href="../regression/kddcup12tr2_lr_amplify.html">
+
+
+ <b>8.3.3.</b>
+
+ Logistic Regression with Amplifier
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.3.4" data-path="../regression/kddcup12tr2_adagrad.html">
+
+ <a href="../regression/kddcup12tr2_adagrad.html">
+
+
+ <b>8.3.4.</b>
+
+ AdaGrad, AdaDelta
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part IX - Recommendation</li>
+
+
+
+ <li class="chapter " data-level="9.1" data-path="cf.html">
+
+ <a href="cf.html">
+
+
+ <b>9.1.</b>
+
+ Collaborative Filtering
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="9.1.1" data-path="item_based_cf.html">
+
+ <a href="item_based_cf.html">
+
+
+ <b>9.1.1.</b>
+
+ Item-based Collaborative Filtering
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="9.2" data-path="news20.html">
+
+ <a href="news20.html">
+
+
+ <b>9.2.</b>
+
+ News20 related article recommendation Tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="9.2.1" data-path="../multiclass/news20_dataset.html">
+
+ <a href="../multiclass/news20_dataset.html">
+
+
+ <b>9.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.2.2" data-path="news20_jaccard.html">
+
+ <a href="news20_jaccard.html">
+
+
+ <b>9.2.2.</b>
+
+ LSH/Minhash and Jaccard Similarity
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.2.3" data-path="news20_knn.html">
+
+ <a href="news20_knn.html">
+
+
+ <b>9.2.3.</b>
+
+ LSH/Minhash and Brute-Force Search
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.2.4" data-path="news20_bbit_minhash.html">
+
+ <a href="news20_bbit_minhash.html">
+
+
+ <b>9.2.4.</b>
+
+ kNN search using b-Bits Minhash
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="9.3" data-path="movielens.html">
+
+ <a href="movielens.html">
+
+
+ <b>9.3.</b>
+
+ MovieLens movie recommendation Tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="9.3.1" data-path="movielens_dataset.html">
+
+ <a href="movielens_dataset.html">
+
+
+ <b>9.3.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.2" data-path="movielens_cf.html">
+
+ <a href="movielens_cf.html">
+
+
+ <b>9.3.2.</b>
+
+ Item-based Collaborative Filtering
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.3" data-path="movielens_mf.html">
+
+ <a href="movielens_mf.html">
+
+
+ <b>9.3.3.</b>
+
+ Matrix Factorization
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.4" data-path="movielens_fm.html">
+
+ <a href="movielens_fm.html">
+
+
+ <b>9.3.4.</b>
+
+ Factorization Machine
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter active" data-level="9.3.5" data-path="movielens_slim.html">
+
+ <a href="movielens_slim.html">
+
+
+ <b>9.3.5.</b>
+
+ SLIM for Fast Top-K Recommendation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.6" data-path="movielens_cv.html">
+
+ <a href="movielens_cv.html">
+
+
+ <b>9.3.6.</b>
+
+ 10-fold Cross Validation (Matrix Factorization)
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part X - Anomaly Detection</li>
+
+
+
+ <li class="chapter " data-level="10.1" data-path="../anomaly/lof.html">
+
+ <a href="../anomaly/lof.html">
+
+
+ <b>10.1.</b>
+
+ Outlier Detection using Local Outlier Factor (LOF)
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="10.2" data-path="../anomaly/sst.html">
+
+ <a href="../anomaly/sst.html">
+
+
+ <b>10.2.</b>
+
+ Change-Point Detection using Singular Spectrum Transformation (SST)
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="10.3" data-path="../anomaly/changefinder.html">
+
+ <a href="../anomaly/changefinder.html">
+
+
+ <b>10.3.</b>
+
+ ChangeFinder: Detecting Outlier and Change-Point Simultaneously
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XI - Clustering</li>
+
+
+
+ <li class="chapter " data-level="11.1" data-path="../clustering/lda.html">
+
+ <a href="../clustering/lda.html">
+
+
+ <b>11.1.</b>
+
+ Latent Dirichlet Allocation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="11.2" data-path="../clustering/plsa.html">
+
+ <a href="../clustering/plsa.html">
+
+
+ <b>11.2.</b>
+
+ Probabilistic Latent Semantic Analysis
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XII - GeoSpatial functions</li>
+
+
+
+ <li class="chapter " data-level="12.1" data-path="../geospatial/latlon.html">
+
+ <a href="../geospatial/latlon.html">
+
+
+ <b>12.1.</b>
+
+ Lat/Lon functions
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XIII - Hivemall on Spark</li>
+
+
+
+ <li class="chapter " data-level="13.1" data-path="../spark/getting_started/">
+
+ <a href="../spark/getting_started/">
+
+
+ <b>13.1.</b>
+
+ Getting Started
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.1.1" data-path="../spark/getting_started/installation.html">
+
+ <a href="../spark/getting_started/installation.html">
+
+
+ <b>13.1.1.</b>
+
+ Installation
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="13.2" data-path="../spark/binaryclass/">
+
+ <a href="../spark/binaryclass/">
+
+
+ <b>13.2.</b>
+
+ Binary Classification
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.2.1" data-path="../spark/binaryclass/a9a_df.html">
+
+ <a href="../spark/binaryclass/a9a_df.html">
+
+
+ <b>13.2.1.</b>
+
+ a9a Tutorial for DataFrame
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="13.2.2" data-path="../spark/binaryclass/a9a_sql.html">
+
+ <a href="../spark/binaryclass/a9a_sql.html">
+
+
+ <b>13.2.2.</b>
+
+ a9a Tutorial for SQL
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="13.3" data-path="../spark/binaryclass/">
+
+ <a href="../spark/binaryclass/">
+
+
+ <b>13.3.</b>
+
+ Regression
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.3.1" data-path="../spark/regression/e2006_df.html">
+
+ <a href="../spark/regression/e2006_df.html">
+
+
+ <b>13.3.1.</b>
+
+ E2006-tfidf regression Tutorial for DataFrame
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="13.3.2" data-path="../spark/regression/e2006_sql.html">
+
+ <a href="../spark/regression/e2006_sql.html">
+
+
+ <b>13.3.2.</b>
+
+ E2006-tfidf regression Tutorial for SQL
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="13.4" data-path="../spark/misc/misc.html">
+
+ <a href="../spark/misc/misc.html">
+
+
+ <b>13.4.</b>
+
+ Generic features
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.4.1" data-path="../spark/misc/topk_join.html">
+
+ <a href="../spark/misc/topk_join.html">
+
+
+ <b>13.4.1.</b>
+
+ Top-k Join processing
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="13.4.2" data-path="../spark/misc/functions.html">
+
+ <a href="../spark/misc/functions.html">
+
+
+ <b>13.4.2.</b>
+
+ Other utility functions
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part XIV - Hivemall on Docker</li>
+
+
+
+ <li class="chapter " data-level="14.1" data-path="../docker/getting_started.html">
+
+ <a href="../docker/getting_started.html">
+
+
+ <b>14.1.</b>
+
+ Getting Started
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XIV - External References</li>
+
+
+
+ <li class="chapter " data-level="15.1" >
+
+ <a target="_blank" href="https://github.com/maropu/hivemall-spark">
+
+
+ <b>15.1.</b>
+
+ Hivemall on Apache Spark
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="15.2" >
+
+ <a target="_blank" href="https://github.com/daijyc/hivemall/wiki/PigHome">
+
+
+ <b>15.2.</b>
+
+ Hivemall on Apache Pig
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="divider"></li>
+
+ <li>
+ <a href="https://www.gitbook.com" target="blank" class="gitbook-link">
+ Published with GitBook
+ </a>
+ </li>
+</ul>
+
+
+ </nav>
+
+
+ </div>
+
+ <div class="book-body">
+
+ <div class="body-inner">
+
+
+
+<div class="book-header" role="navigation">
+
+
+ <!-- Title -->
+ <h1>
+ <i class="fa fa-circle-o-notch fa-spin"></i>
+ <a href=".." >SLIM for Fast Top-K Recommendation</a>
+ </h1>
+</div>
+
+
+
+
+ <div class="page-wrapper" tabindex="-1" role="main">
+ <div class="page-inner">
+
+<div id="book-search-results">
+ <div class="search-noresults">
+
+ <section class="normal markdown-section">
+
+ <!--
+ Licensed to the Apache Software Foundation (ASF) under one
+ or more contributor license agreements. See the NOTICE file
+ distributed with this work for additional information
+ regarding copyright ownership. The ASF licenses this file
+ to you under the Apache License, Version 2.0 (the
+ "License"); you may not use this file except in compliance
+ with the License. You may obtain a copy of the License at
+
+ http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing,
+ software distributed under the License is distributed on an
+ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ KIND, either express or implied. See the License for the
+ specific language governing permissions and limitations
+ under the License.
+-->
+<p>Hivemall supports a neighborhood-learning scheme using SLIM.
+SLIM is a representative of neighborhood-learning recommendation algorithm introduced in the following paper:</p>
+<ul>
+<li>Xia Ning and George Karypis, <a href="https://dl.acm.org/citation.cfm?id=2118303" target="_blank">SLIM: Sparse Linear Methods for Top-N Recommender Systems</a>, Proc. ICDM, 2011.</li>
+</ul>
+<p><em>Caution: SLIM is supported from Hivemall v0.5-rc.1 or later.</em></p>
+<!-- toc --><div id="toc" class="toc">
+
+<ul>
+<li><a href="#slim-optimization-objective">SLIM optimization objective</a></li>
+<li><a href="#data-preparation">Data preparation</a><ul>
+<li><a href="#rating-binarization">Rating binarization</a></li>
+<li><a href="#splitting-dataset">Splitting dataset</a><ul>
+<li><a href="#leave-one-out-cross-validation">Leave-one-out cross validation</a></li>
+<li><a href="#k-hold-corss-validation"><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi></mrow><annotation encoding="application/x-tex">K</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span></span></span></span>-hold corss validation</a></li>
+</ul>
+</li>
+<li><a href="#precompute-movie-movie-similarity">Precompute movie-movie similarity</a></li>
+<li><a href="#create-training-input-tables">Create training input tables</a></li>
+</ul>
+</li>
+<li><a href="#training">Training</a><ul>
+<li><a href="#build-a-prediction-model-by-slim">Build a prediction model by SLIM</a></li>
+<li><a href="#usage-of-trainslim">Usage of <code>train_slim</code></a></li>
+</ul>
+</li>
+<li><a href="#prediction-and-recommendation">Prediction and recommendation</a><ul>
+<li><a href="#predict-unknown-value-of-user-item-matrix">Predict unknown value of user-item matrix</a></li>
+</ul>
+</li>
+<li><a href="#evaluation">Evaluation</a><ul>
+<li><a href="#top-k-ranking-measures-hit-ratek-mrrk-and-precisionk">Top-<span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi></mrow><annotation encoding="application/x-tex">K</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span></span></span></span> ranking measures: Hit-Rate@K, MRR@K, and Precision@K</a><ul>
+<li><a href="#leave-one-out-result">Leave-one-out result</a></li>
+<li><a href="#k-hold-result"><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi></mrow><annotation encoding="application/x-tex">K</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span></span></span></span>-hold result</a></li>
+</ul>
+</li>
+<li><a href="#ranking-measures-mrr">Ranking measures: MRR</a><ul>
+<li><a href="#leave-one-out-result-1">Leave-one-out result</a></li>
+</ul>
+</li>
+</ul>
+</li>
+</ul>
+
+</div><!-- tocstop -->
+<h1 id="slim-optimization-objective">SLIM optimization objective</h1>
+<p>The optimization objective of <a href="(http:/glaros.dtc.umn.edu/gkhome/fetch/papers/SLIM2011icdm.pdf">SLIM</a>) is similar to Elastic Net (L1+L2 regularization) with additional constraints as follows:</p>
+<p><span class="katex-display"><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mtable><mtr><mtd><mrow></mrow></mtd><mtd><mrow><mrow></mrow><mspace width="0.277778em"></mspace><mrow><mstyle mathsize="0.5em"><mtable><mtr><mtd><mrow></mrow></mtd></mtr><mtr><mtd><mrow><mstyle mathsize="1em"><mtext><mi mathvariant="normal">m</mi><mi mathvariant="normal">i</mi><mi mathvariant="normal">n</mi><mi mathvariant="normal">i</mi><mi mathvariant="normal">m</mi><mi mathvariant="normal">i</mi><mi mathvariant="normal">z</mi><mi mathvariant="normal">e</mi></mtext></mstyle></mrow></mtd></mtr><mtr><mtd><mrow><msup><mrow></mrow><mrow><mstyle mathsize="0.7em"><msub><mi>w</mi><mrow><mi>j</mi></mrow></msub></mstyle></mrow></msup></mrow></mtd></mtr></mtable></mstyle></mrow><mspace width="0.277778em"></mspace></mrow></mtd><mtd><mrow></mrow></mtd><mtd><mrow><mrow></mrow><mfrac><mrow><mn>1</mn></mrow><mrow><mn>2</mn></mrow></mfrac><mi mathvariant="normal">∥</mi><msub><mi>r</mi><mro
w><mi>j</mi></mrow></msub><mo>−</mo><mi>R</mi><msub><mi>w</mi><mrow><mi>j</mi></mrow></msub><msubsup><mi mathvariant="normal">∥</mi><mn>2</mn><mn>2</mn></msubsup><mo>+</mo><mfrac><mrow><mi>β</mi></mrow><mrow><mn>2</mn></mrow></mfrac><mi mathvariant="normal">∥</mi><msub><mi>w</mi><mrow><mi>j</mi></mrow></msub><msubsup><mi mathvariant="normal">∥</mi><mn>2</mn><mn>2</mn></msubsup><mo>+</mo><mi>λ</mi><mi mathvariant="normal">∥</mi><msub><mi>w</mi><mrow><mi>j</mi></mrow></msub><msub><mi mathvariant="normal">∥</mi><mn>1</mn></msub></mrow></mtd></mtr><mtr><mtd><mrow></mrow></mtd><mtd><mrow><mrow></mrow><mtext><mi mathvariant="normal">s</mi><mi mathvariant="normal">u</mi><mi mathvariant="normal">b</mi><mi mathvariant="normal">j</mi><mi mathvariant="normal">e</mi><mi mathvariant="normal">c</mi><mi mathvariant="normal">t</mi><mtext> </mtext><mi mathvariant="normal">t</mi><mi mathvariant="normal">o</mi></mtext></mrow></mtd><mtd><mrow><
/mrow></mtd><mtd><mrow><mrow></mrow><msub><mi>w</mi><mrow><mi>j</mi></mrow></msub><mo>≥</mo><mn>0</mn></mrow></mtd></mtr><mtr><mtd><mrow></mrow></mtd><mtd><mrow><mrow></mrow></mrow></mtd><mtd><mrow></mrow></mtd><mtd><mrow><mrow></mrow><mi>d</mi><mi>i</mi><mi>a</mi><mi>g</mi><mo>(</mo><mi>W</mi><mo>)</mo><mo>=</mo><mn>0</mn></mrow></mtd></mtr></mtable></mrow><annotation encoding="application/x-tex">
+\begin{aligned}
+& \;{\tiny\begin{matrix}\\ \normalsize \text{minimize} \\ ^{\scriptsize w_{j}}\end{matrix}}\;
+&& \frac{1}{2}\Vert r_{j} - Rw_{j} \Vert_2^2 + \frac{\beta}{2} \Vert w_{j} \Vert_2^2 + \lambda \Vert w_{j} \Vert_1 \\
+& \text{subject to}
+&& w_{j} \geq 0 \\
+&&& diag(W)= 0
+\end{aligned}
+</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:2.5232200000000002em;"></span><span class="strut bottom" style="height:4.5464400000000005em;vertical-align:-2.02322em;"></span><span class="base displaystyle textstyle uncramped"><span class="mord"><span class="mtable"><span class="col-align-r"><span class="vlist"><span style="top:-1.1517800000000002em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></span><span style="top:0.4632199999999996em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></span><span style="top:1.6632199999999997em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></s
pan><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span><span class="col-align-l"><span class="vlist"><span style="top:-1.1517800000000002em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0.5em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord displaystyle textstyle uncramped"></span><span class="mord displaystyle textstyle uncramped"><span class="mspace thickspace"></span><span class="mord sizing reset-size5 size1 displaystyle textstyle uncramped"><span class="mtable"><span class="col-align-c"><span class="vlist"><span style="top:-1.2099999999999997em;"><span class="fontsize-ensurer reset-size1 size5"><span style="font-size:1em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></span><span style="top:-0.00999999999999951em;"><span class="fontsize-ensurer reset-size1 size5"><spa
n style="font-size:1em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord text displaystyle textstyle uncramped sizing reset-size1 size5 displaystyle textstyle uncramped"><span class="mord mathrm">minimize</span></span></span></span><span style="top:1.1900000000000006em;"><span class="fontsize-ensurer reset-size1 size5"><span style="font-size:1em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord"><span></span><span class="msupsub"><span class="vlist"><span style="top:-0.413em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size1 size5"><span style="font-size:0.48999999999999994em;">​</span></span><span class="reset-textstyle scriptstyle uncramped mtight"><span class="mord scriptstyle uncramped mtight"><span class="mord mtight sizing reset-size1 size2 scriptstyle uncramped"><span class="mord mathit mtight" style="margin-right:0.02691em;">w</span><span class="msupsub"><span class="v
list"><span style="top:0.14300000000000002em;margin-right:0.07142857142857144em;margin-left:-0.02691em;"><span class="fontsize-ensurer reset-size2 size5"><span style="font-size:0em;">​</span></span><span class="reset-scriptstyle scriptscriptstyle cramped mtight"><span class="mord scriptscriptstyle cramped mtight"><span class="mord mathit mtight" style="margin-right:0.05724em;">j</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size2 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size1 size5"><span style="font-size:0.48999999999999994em;">​</span></span>​</span></span></span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size1 size5"><span style="font-size:1em;">​</span></span>​</span></span></span></span></span></span><span class="mspace thick
space"></span></span></span><span style="top:0.4632199999999996em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0.5em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord displaystyle textstyle uncramped"></span><span class="mord text displaystyle textstyle uncramped"><span class="mord mathrm">subject to</span></span></span></span><span style="top:1.6632199999999997em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0.5em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord displaystyle textstyle uncramped"></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0.5em;">​</span></span>​</span></span></span><span class="arraycolsep" style="width:2em;"></span><span class="col-align-r"><span class="vlist"><span style="top:-1.1517800000000002em;"><span class="fontsize-ensu
rer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></span><span style="top:0.4632199999999996em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></span><span style="top:1.6632199999999997em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span><span class="col-align-l"><span class="vlist"><span style="top:-1.1517800000000002em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord displaystyle textstyle uncramped
"></span><span class="mord reset-textstyle displaystyle textstyle uncramped"><span class="mopen sizing reset-size5 size5 reset-textstyle textstyle uncramped nulldelimiter"></span><span class="mfrac"><span class="vlist"><span style="top:0.686em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle cramped"><span class="mord textstyle cramped"><span class="mord mathrm">2</span></span></span></span><span style="top:-0.22999999999999998em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle uncramped frac-line"></span></span><span style="top:-0.677em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathrm">1</span></span></span></span><span class="baseline-f
ix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span><span class="mclose sizing reset-size5 size5 reset-textstyle textstyle uncramped nulldelimiter"></span></span><span class="mord mathrm">∥</span><span class="mord"><span class="mord mathit" style="margin-right:0.02778em;">r</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.02778em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord scriptstyle cramped mtight"><span class="mord mathit mtight" style="margin-right:0.05724em;">j</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mbin">−</span><span class="mord mathi
t" style="margin-right:0.00773em;">R</span><span class="mord"><span class="mord mathit" style="margin-right:0.02691em;">w</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.02691em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord scriptstyle cramped mtight"><span class="mord mathit mtight" style="margin-right:0.05724em;">j</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mord"><span class="mord mathrm">∥</span><span class="msupsub"><span class="vlist"><span style="top:0.247em;margin-left:0em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scri
ptstyle cramped mtight"><span class="mord mathrm mtight">2</span></span></span><span style="top:-0.4129999999999999em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle uncramped mtight"><span class="mord mathrm mtight">2</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mbin">+</span><span class="mord reset-textstyle displaystyle textstyle uncramped"><span class="mopen sizing reset-size5 size5 reset-textstyle textstyle uncramped nulldelimiter"></span><span class="mfrac"><span class="vlist"><span style="top:0.686em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle cramped"><span class="mord textstyle cramped"><span class="mord mathrm">2</sp
an></span></span></span><span style="top:-0.22999999999999998em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle uncramped frac-line"></span></span><span style="top:-0.677em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathit" style="margin-right:0.05278em;">β</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span><span class="mclose sizing reset-size5 size5 reset-textstyle textstyle uncramped nulldelimiter"></span></span><span class="mord mathrm">∥</span><span class="mord"><span class="mord mathit" style="margin-right:0.02691em;">w</span><span class="msupsub"><span class="vlist"><span style="
top:0.15em;margin-right:0.05em;margin-left:-0.02691em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord scriptstyle cramped mtight"><span class="mord mathit mtight" style="margin-right:0.05724em;">j</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mord"><span class="mord mathrm">∥</span><span class="msupsub"><span class="vlist"><span style="top:0.247em;margin-left:0em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathrm mtight">2</span></span></span><span style="top:-0.4129999999999999em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 s
ize5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle uncramped mtight"><span class="mord mathrm mtight">2</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mbin">+</span><span class="mord mathit">λ</span><span class="mord mathrm">∥</span><span class="mord"><span class="mord mathit" style="margin-right:0.02691em;">w</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.02691em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord scriptstyle cramped mtight"><span class="mord mathit mtight" style="margin-right:0.05724em;">j</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset
-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mord"><span class="mord mathrm">∥</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:0em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathrm mtight">1</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span></span></span><span style="top:0.4632199999999996em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord displaystyle textstyle uncramped"></span><span class="mord"><span class="mord mathit" style="margin-right:0.02691em;">w</span><span c
lass="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.02691em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord scriptstyle cramped mtight"><span class="mord mathit mtight" style="margin-right:0.05724em;">j</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mrel">≥</span><span class="mord mathrm">0</span></span></span><span style="top:1.6632199999999997em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="mord displaystyle textstyle uncramped"><span class="mord displaystyle textstyle uncramped"></span><span class="mord mathit">d</span><span class="mord mathit">i</span><span class="mord mathit">a</spa
n><span class="mord mathit" style="margin-right:0.03588em;">g</span><span class="mopen">(</span><span class="mord mathit" style="margin-right:0.13889em;">W</span><span class="mclose">)</span><span class="mrel">=</span><span class="mord mathrm">0</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span></span></span></span></span></span></p>
+<h1 id="data-preparation">Data preparation</h1>
+<h2 id="rating-binarization">Rating binarization</h2>
+<p>In this article, each user-movie matrix element is binarized to reduce training samples and consider only high rated movies whose rating is 4 or 5. So, every matrix element having a lower rating than 4 is not used for training.</p>
+<pre><code class="lang-sql"><span class="hljs-keyword">SET</span> hivevar:<span class="hljs-keyword">seed</span>=<span class="hljs-number">31</span>;
+
+<span class="hljs-keyword">DROP</span> <span class="hljs-keyword">TABLE</span> ratings2;
+<span class="hljs-keyword">CREATE</span> <span class="hljs-keyword">TABLE</span> ratings2 <span class="hljs-keyword">as</span>
+<span class="hljs-keyword">select</span>
+ <span class="hljs-keyword">rand</span>(${<span class="hljs-keyword">seed</span>}) <span class="hljs-keyword">as</span> rnd,
+ userid,
+ movieid <span class="hljs-keyword">as</span> itemid,
+ <span class="hljs-keyword">cast</span>(<span class="hljs-number">1.0</span> <span class="hljs-keyword">as</span> <span class="hljs-built_in">float</span>) <span class="hljs-keyword">as</span> rating <span class="hljs-comment">-- double is also accepted</span>
+<span class="hljs-keyword">from</span>
+ ratings
+<span class="hljs-keyword">where</span> rating >= <span class="hljs-number">4.</span>
+;
+</code></pre>
+<p><code>rnd</code> field is appended for each record to split <code>ratings2</code> into training and testing data later.</p>
+<p>Binarization is an optional step, and you can use raw rating values to train a SLIM model.</p>
+<h2 id="splitting-dataset">Splitting dataset</h2>
+<p>To evaluate a recommendation model, this tutorial uses two type cross validations:</p>
+<ul>
+<li>Leave-one-out cross validation</li>
+<li><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi></mrow><annotation encoding="application/x-tex">K</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span></span></span></span>-hold cross validation</li>
+</ul>
+<p>The former is used in the <a href="http://glaros.dtc.umn.edu/gkhome/fetch/papers/SLIM2011icdm.pdf" target="_blank">SLIM's paper</a> and the latter is used in <a href="http://slideshare.net/MarkLevy/efficient-slides/" target="_blank">Mendeley's slide</a>.</p>
+<h3 id="leave-one-out-cross-validation">Leave-one-out cross validation</h3>
+<p>For leave-one-out cross validation, the dataset is split into a training set and a testing set by randomly selecting one of the non-zero entries of each user and placing it into the testing set.
+In the following query, the movie has the smallest <code>rnd</code> value is used as test data (<code>testing</code> table) per a user.
+And, the others are used as training data (<code>training</code> table).</p>
+<p>When we select slim's best hyperparameters, different test data is used in <a href="#evaluation">evaluation section</a> several times.</p>
+<pre><code class="lang-sql"><span class="hljs-keyword">DROP</span> <span class="hljs-keyword">TABLE</span> testing;
+<span class="hljs-keyword">CREATE</span> <span class="hljs-keyword">TABLE</span> testing
+<span class="hljs-keyword">as</span>
+<span class="hljs-keyword">WITH</span> top_k <span class="hljs-keyword">as</span> (
+ <span class="hljs-keyword">select</span>
+ each_top_k(<span class="hljs-number">1</span>, userid, rnd, userid, itemid, rating)
+ <span class="hljs-keyword">as</span> (<span class="hljs-keyword">rank</span>, rnd, userid, itemid, rating)
+ <span class="hljs-keyword">from</span> (
+ <span class="hljs-keyword">select</span> * <span class="hljs-keyword">from</span> ratings2
+ CLUSTER <span class="hljs-keyword">BY</span> userid
+ ) t
+)
+<span class="hljs-keyword">select</span>
+ userid, itemid, rating
+<span class="hljs-keyword">from</span>
+ top_k
+;
+
+<span class="hljs-keyword">DROP</span> <span class="hljs-keyword">TABLE</span> training;
+<span class="hljs-keyword">CREATE</span> <span class="hljs-keyword">TABLE</span> training <span class="hljs-keyword">as</span>
+<span class="hljs-keyword">select</span>
+ l.*
+<span class="hljs-keyword">from</span>
+ ratings2 l
+ <span class="hljs-keyword">LEFT</span> <span class="hljs-keyword">OUTER</span> <span class="hljs-keyword">JOIN</span> testing r <span class="hljs-keyword">ON</span> (l.userid=r.userid <span class="hljs-keyword">and</span> l.itemid=r.itemid)
+<span class="hljs-keyword">where</span>
+ r.itemid <span class="hljs-keyword">IS</span> <span class="hljs-literal">NULL</span> <span class="hljs-comment">-- anti join</span>
+;
+</code></pre>
+<h3 id="kkk-hold-corss-validation"><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi></mrow><annotation encoding="application/x-tex">K</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span></span></span></span>-hold corss validation</h3>
+<p>When <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi><mo>=</mo><mn>2</mn></mrow><annotation encoding="application/x-tex">K=2</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span><span class="mrel">=</span><span class="mord mathrm">2</span></span></span></span>, the dataset is divided into training data and testing dataset.
+The numbers of training and testing samples roughly equal.</p>
+<p>When we select slim's best hyperparameters, you'll first train a SLIM prediction model from training data and evaluate the prediction model by testing data.</p>
+<p>Optionally, you can switch training data with testing data and evaluate again.</p>
+<pre><code class="lang-sql"><span class="hljs-keyword">DROP</span> <span class="hljs-keyword">TABLE</span> testing;
+<span class="hljs-keyword">CREATE</span> <span class="hljs-keyword">TABLE</span> testing
+<span class="hljs-keyword">as</span>
+<span class="hljs-keyword">select</span> * <span class="hljs-keyword">from</span> ratings2
+<span class="hljs-keyword">where</span> rnd >= <span class="hljs-number">0.5</span>
+;
+
+<span class="hljs-keyword">DROP</span> <span class="hljs-keyword">TABLE</span> training;
+<span class="hljs-keyword">CREATE</span> <span class="hljs-keyword">TABLE</span> training
+<span class="hljs-keyword">as</span>
+<span class="hljs-keyword">select</span> * <span class="hljs-keyword">from</span> ratings2
+<span class="hljs-keyword">where</span> rnd < <span class="hljs-number">0.5</span>
+;
+</code></pre>
+<div class="panel panel-primary"><div class="panel-heading"><h3 class="panel-title" id="note"><i class="fa fa-edit"></i> Note</h3></div><div class="panel-body"><p>In the following section excluding evaluation section,
+we will show the example of queries and its results based on <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>K</mi></mrow><annotation encoding="application/x-tex">K</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.68333em;"></span><span class="strut bottom" style="height:0.68333em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.07153em;">K</span></span></span></span>-hold cross validation case.
+But, this article's queries are valid for leave-one-out cross validation.</p></div></div>
+<h2 id="precompute-movie-movie-similarity">Precompute movie-movie similarity</h2>
+<p>SLIM needs top-<span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>k</mi></mrow><annotation encoding="application/x-tex">k</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.69444em;"></span><span class="strut bottom" style="height:0.69444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.03148em;">k</span></span></span></span> most similar movies for each movie to the approximate user-item matrix.
+Here, we particularly focus on <a href="item_based_cf.html#dimsum-approximated-all-pairs-cosine-similarity-computation">DIMSUM</a>,
+an efficient and approximated similarity computation scheme.</p>
+<p>Because we set <code>k=20</code>, the output has 20 most-similar movies per <code>itemid</code>.
+We can adjust trade-off between training and prediction time and precision of matrix approximation by varying <code>k</code>.
+Larger <code>k</code> is the better approximation for raw user-item matrix, but training time and memory usage tend to increase.</p>
+<p><a href="item_based_cf.html#dimsum-approximated-all-pairs-cosine-similarity-computation.md">As we explained in the general introduction of item-based CF</a>,
+following query finds top-<span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>k</mi></mrow><annotation encoding="application/x-tex">k</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.69444em;"></span><span class="strut bottom" style="height:0.69444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.03148em;">k</span></span></span></span> nearest-neighborhood movies for each movie:</p>
+<pre><code class="lang-sql"><span class="hljs-keyword">set</span> hivevar:k=<span class="hljs-number">20</span>;
+
+<span class="hljs-keyword">DROP</span> <span class="hljs-keyword">TABLE</span> knn_train;
+<span class="hljs-keyword">CREATE</span> <span class="hljs-keyword">TABLE</span> knn_train
+<span class="hljs-keyword">as</span>
+<span class="hljs-keyword">with</span> item_magnitude <span class="hljs-keyword">as</span> (
+ <span class="hljs-keyword">select</span>
+ to_map(j, mag) <span class="hljs-keyword">as</span> mags
+ <span class="hljs-keyword">from</span> (
+ <span class="hljs-keyword">select</span>
+ itemid <span class="hljs-keyword">as</span> j,
+ l2_norm(rating) <span class="hljs-keyword">as</span> mag
+ <span class="hljs-keyword">from</span>
+ training
+ <span class="hljs-keyword">group</span> <span class="hljs-keyword">by</span>
+ itemid
+ ) t0
+),
+item_features <span class="hljs-keyword">as</span> (
+ <span class="hljs-keyword">select</span>
+ userid <span class="hljs-keyword">as</span> i,
+ collect_list(
+ feature(itemid, rating)
+ ) <span class="hljs-keyword">as</span> feature_vector
+ <span class="hljs-keyword">from</span>
+ training
+ <span class="hljs-keyword">group</span> <span class="hljs-keyword">by</span>
+ userid
+),
+partial_result <span class="hljs-keyword">as</span> (
+ <span class="hljs-keyword">select</span>
+ dimsum_mapper(f.feature_vector, m.mags, <span class="hljs-string">'-threshold 0.1 -int_feature'</span>)
+ <span class="hljs-keyword">as</span> (itemid, other, s)
+ <span class="hljs-keyword">from</span>
+ item_features f
+ <span class="hljs-keyword">CROSS</span> <span class="hljs-keyword">JOIN</span> item_magnitude m
+),
+similarity <span class="hljs-keyword">as</span> (
+ <span class="hljs-keyword">select</span>
+ itemid,
+ other,
+ <span class="hljs-keyword">sum</span>(s) <span class="hljs-keyword">as</span> similarity
+ <span class="hljs-keyword">from</span>
+
<TRUNCATED>