Update Movie Recommendations Notebook

Author: ksalama

Experimental/Movielens Recommendation.ipynb

@@ -1,5 +1,45 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Recommendation Model with Approximate Item Matching\n",
+    "\n",
+    "This notebook shows how to train a simple Neural Collaborative Filtering model for recommeding movies to users. We also show how learnt movie embeddings are stored in an appoximate similarity matching index, using Spotify's [Annoy library](https://github.com/spotify/annoy), so that we can quickly find and recommend the most relevant movies to a given customer. We show how this index to search for similar movies.\n",
+    "\n",
+    "In essense, this tutorial works as follows:\n",
+    "1. Download the movielens dataset.\n",
+    "2. Train a simple Neural Collaborative Model using TensorFlow custom estimator.\n",
+    "3. Extract the learnt movie embeddings.\n",
+    "4. Build an approximate similarity matching index for the movie embeddings.\n",
+    "5. Export the trained model, which receives a user Id, and output the user embedding.\n",
+    "\n",
+    "The recommendation is served as follows:\n",
+    "1. Receives a user Id\n",
+    "2. Get the user embedding from the exported model\n",
+    "3. Find the similar movie embeddings to the user embedding in the index\n",
+    "4. Return the movie Ids of these embeddings to recommend\n",
+    "\n",
+    "<a href=\"https://colab.research.google.com/github/GoogleCloudPlatform/tf-estimator-tutorials/blob/master/Experimental/Movielens%20Recommendation.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Setup"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install annoy"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 1,
@@ -32,7 +72,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Download Data"
+    "## 1. Download Data"
    ]
   },
   {
@@ -373,7 +413,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Define Metadata"
+    "## 2. Build the TensorFlow Model"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2.1 Define Metadata"
    ]
   },
   {
@@ -393,12 +440,12 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Define Data Input Function"
+    "### 2.2 Define Data Input Function"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 18,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -418,10 +465,7 @@
     "            num_epochs=num_epochs,\n",
     "            shuffle= (mode==tf.estimator.ModeKeys.TRAIN)\n",
     "        )\n",
-    "        \n",
-    "        iterator = dataset.make_one_shot_iterator()\n",
-    "        features, target = iterator.get_next()\n",
-    "        return features, target\n",
+    "        return dataset\n",
     "    \n",
     "    return _input_fn"
    ]
@@ -430,7 +474,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Create Feature Columns"
+    "### 2.3 Create Feature Columns"
    ]
   },
   {
@@ -466,7 +510,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Define Model Function"
+    "### 2.4 Define Model Function"
    ]
   },
   {
@@ -506,14 +550,14 @@
     "        mode=mode,\n",
     "        loss=loss,\n",
     "        train_op=train_op\n",
-    "    )\n"
+    "    )"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Create Estimator"
+    "### 2.5 Create Estimator"
    ]
   },
   {
@@ -537,7 +581,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Define Experiment"
+    "### 2.6 Define Experiment"
    ]
   },
   {
@@ -612,7 +656,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Run Experiment with Parameters"
+    "### 2.7 Run Experiment with Parameters"
    ]
   },
   {
@@ -710,7 +754,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Extract Movie Embeddings "
+    "## 3. Extract Movie Embeddings "
    ]
   },
   {
@@ -766,7 +810,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Build Annoy Index"
+    "## 4. Build Annoy Index"
    ]
   },
   {
@@ -1145,7 +1189,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Export the Model\n",
+    "## 5. Export the Model\n",
     "This needed to receive a userId and produce the embedding for the user."
    ]
   },
@@ -1234,6 +1278,13 @@
     "print(output)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Serve Movie Recommendations to a User"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 190,
@@ -1276,11 +1327,43 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": null,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
-   "source": []
+   "source": [
+    "## License"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "---\n",
+    "\n",
+    "Author: Khalid Salama\n",
+    "\n",
+    "\n",
+    "---\n",
+    "***Disclaimer***: This is not an official Google product. This sample code provided for an educational purpose.\n",
+    "\n",
+    "---\n",
+    "\n",
+    "Copyright 2019 Google LLC\n",
+    "\n",
+    "Licensed under the Apache License, Version 2.0 (the \"License\");\n",
+    "you may not use this file except in compliance with the License.\n",
+    "You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0.\n",
+    "\n",
+    "Unless required by applicable law or agreed to in writing, software\n",
+    "distributed under the License is distributed on an \"AS IS\" BASIS,\n",
+    "WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
+    "See the License for the specific language governing permissions and\n",
+    "limitations under the License.\n",
+    "\n",
+    "\n",
+    "---\n",
+    "\n",
+    "\n"
+   ]
   }
  ],
  "metadata": {