ondisk_dataset_heterograph.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "private_outputs": true,
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# OnDiskDataset for Heterogeneous Graph\n",
        "\n",
        "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/dmlc/dgl/blob/master/notebooks/stochastic_training/ondisk_dataset_heterograph.ipynb) [![GitHub](https://img.shields.io/badge/-View%20on%20GitHub-181717?logo=github&logoColor=ffffff)](https://github.com/dmlc/dgl/blob/master/notebooks/stochastic_training/ondisk_dataset_heterograph.ipynb)\n",
        "\n",
        "This tutorial shows how to create `OnDiskDataset` for heterogeneous graph that could be used in **GraphBolt** framework. The major difference from creating dataset for homogeneous graph is that we need to specify node/edge types for edges, feature data, training/validation/test sets.\n",
        "\n",
        "By the end of this tutorial, you will be able to\n",
        "- organize graph structure data.\n",
        "- organize feature data.\n",
        "- organize training/validation/test set for specific tasks.\n",
        "\n",
        "To create an ``OnDiskDataset`` object, you need to organize all the data including graph structure, feature data and tasks into a directory. The directory should contain a ``metadata.yaml`` file that describes the metadata of the dataset.\n",
        "\n",
        "Now let's generate various data step by step and organize them together to instantiate `OnDiskDataset` finally."
      ],
      "metadata": {
        "id": "FnFhPMaAfLtJ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Install DGL package"
      ],
      "metadata": {
        "id": "Wlb19DtWgtzq"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Install required packages.\n",
        "import os\n",
        "import torch\n",
        "import numpy as np\n",
        "os.environ['TORCH'] = torch.__version__\n",
        "os.environ['DGLBACKEND'] = \"pytorch\"\n",
        "\n",
        "# Install the CPU version.\n",
        "device = torch.device(\"cpu\")\n",
        "!pip install --pre dgl -f https://data.dgl.ai/wheels-test/repo.html\n",
        "\n",
        "try:\n",
        "    import dgl\n",
        "    import dgl.graphbolt as gb\n",
        "    installed = True\n",
        "except ImportError as error:\n",
        "    installed = False\n",
        "    print(error)\n",
        "print(\"DGL installed!\" if installed else \"DGL not found!\")"
      ],
      "metadata": {
        "id": "UojlT9ZGgyr9"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Data preparation\n",
        "In order to demonstrate how to organize various data, let's create a base directory first."
      ],
      "metadata": {
        "id": "2R7WnSbjsfbr"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "base_dir = './ondisk_dataset_heterograph'\n",
        "os.makedirs(base_dir, exist_ok=True)\n",
        "print(f\"Created base directory: {base_dir}\")"
      ],
      "metadata": {
        "id": "SZipbzyltLfO"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Generate graph structure data\n",
        "For heterogeneous graph, we need to save different edge edges(namely node pairs) into separate **CSV** files.\n",
        "\n",
        "Note:\n",
        "when saving to file, do not save index and header.\n"
      ],
      "metadata": {
        "id": "qhNtIn_xhlnl"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "\n",
        "# For simplicity, we create a heterogeneous graph with\n",
        "# 2 node types: `user`, `item`\n",
        "# 2 edge types: `user:like:item`, `user:follow:user`\n",
        "# And each node/edge type has the same number of nodes/edges.\n",
        "num_nodes = 1000\n",
        "num_edges = 10 * num_nodes\n",
        "\n",
        "# Edge type: \"user:like:item\"\n",
        "like_edges_path = os.path.join(base_dir, \"like-edges.csv\")\n",
        "like_edges = np.random.randint(0, num_nodes, size=(num_edges, 2))\n",
        "print(f\"Part of [user:like:item] edges: {like_edges[:10, :]}\\n\")\n",
        "\n",
        "df = pd.DataFrame(like_edges)\n",
        "df.to_csv(like_edges_path, index=False, header=False)\n",
        "print(f\"[user:like:item] edges are saved into {like_edges_path}\\n\")\n",
        "\n",
        "# Edge type: \"user:follow:user\"\n",
        "follow_edges_path = os.path.join(base_dir, \"follow-edges.csv\")\n",
        "follow_edges = np.random.randint(0, num_nodes, size=(num_edges, 2))\n",
        "print(f\"Part of [user:follow:user] edges: {follow_edges[:10, :]}\\n\")\n",
        "\n",
        "df = pd.DataFrame(follow_edges)\n",
        "df.to_csv(follow_edges_path, index=False, header=False)\n",
        "print(f\"[user:follow:user] edges are saved into {follow_edges_path}\\n\")"
      ],
      "metadata": {
        "id": "HcBt4G5BmSjr"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Generate feature data for graph\n",
        "For feature data, numpy arrays and torch tensors are supported for now. Let's generate feature data for each node/edge type."
      ],
      "metadata": {
        "id": "kh-4cPtzpcaH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Generate node[user] feature in numpy array.\n",
        "node_user_feat_0_path = os.path.join(base_dir, \"node-user-feat-0.npy\")\n",
        "node_user_feat_0 = np.random.rand(num_nodes, 5)\n",
        "print(f\"Part of node[user] feature [feat_0]: {node_user_feat_0[:10, :]}\")\n",
        "np.save(node_user_feat_0_path, node_user_feat_0)\n",
        "print(f\"Node[user] feature [feat_0] is saved to {node_user_feat_0_path}\\n\")\n",
        "\n",
        "# Generate another node[user] feature in torch tensor\n",
        "node_user_feat_1_path = os.path.join(base_dir, \"node-user-feat-1.pt\")\n",
        "node_user_feat_1 = torch.rand(num_nodes, 5)\n",
        "print(f\"Part of node[user] feature [feat_1]: {node_user_feat_1[:10, :]}\")\n",
        "torch.save(node_user_feat_1, node_user_feat_1_path)\n",
        "print(f\"Node[user] feature [feat_1] is saved to {node_user_feat_1_path}\\n\")\n",
        "\n",
        "# Generate node[item] feature in numpy array.\n",
        "node_item_feat_0_path = os.path.join(base_dir, \"node-item-feat-0.npy\")\n",
        "node_item_feat_0 = np.random.rand(num_nodes, 5)\n",
        "print(f\"Part of node[item] feature [feat_0]: {node_item_feat_0[:10, :]}\")\n",
        "np.save(node_item_feat_0_path, node_item_feat_0)\n",
        "print(f\"Node[item] feature [feat_0] is saved to {node_item_feat_0_path}\\n\")\n",
        "\n",
        "# Generate another node[item] feature in torch tensor\n",
        "node_item_feat_1_path = os.path.join(base_dir, \"node-item-feat-1.pt\")\n",
        "node_item_feat_1 = torch.rand(num_nodes, 5)\n",
        "print(f\"Part of node[item] feature [feat_1]: {node_item_feat_1[:10, :]}\")\n",
        "torch.save(node_item_feat_1, node_item_feat_1_path)\n",
        "print(f\"Node[item] feature [feat_1] is saved to {node_item_feat_1_path}\\n\")\n",
        "\n",
        "# Generate edge[user:like:item] feature in numpy array.\n",
        "edge_like_feat_0_path = os.path.join(base_dir, \"edge-like-feat-0.npy\")\n",
        "edge_like_feat_0 = np.random.rand(num_edges, 5)\n",
        "print(f\"Part of edge[user:like:item] feature [feat_0]: {edge_like_feat_0[:10, :]}\")\n",
        "np.save(edge_like_feat_0_path, edge_like_feat_0)\n",
        "print(f\"Edge[user:like:item] feature [feat_0] is saved to {edge_like_feat_0_path}\\n\")\n",
        "\n",
        "# Generate another edge[user:like:item] feature in torch tensor\n",
        "edge_like_feat_1_path = os.path.join(base_dir, \"edge-like-feat-1.pt\")\n",
        "edge_like_feat_1 = torch.rand(num_edges, 5)\n",
        "print(f\"Part of edge[user:like:item] feature [feat_1]: {edge_like_feat_1[:10, :]}\")\n",
        "torch.save(edge_like_feat_1, edge_like_feat_1_path)\n",
        "print(f\"Edge[user:like:item] feature [feat_1] is saved to {edge_like_feat_1_path}\\n\")\n",
        "\n",
        "# Generate edge[user:follow:user] feature in numpy array.\n",
        "edge_follow_feat_0_path = os.path.join(base_dir, \"edge-follow-feat-0.npy\")\n",
        "edge_follow_feat_0 = np.random.rand(num_edges, 5)\n",
        "print(f\"Part of edge[user:follow:user] feature [feat_0]: {edge_follow_feat_0[:10, :]}\")\n",
        "np.save(edge_follow_feat_0_path, edge_follow_feat_0)\n",
        "print(f\"Edge[user:follow:user] feature [feat_0] is saved to {edge_follow_feat_0_path}\\n\")\n",
        "\n",
        "# Generate another edge[user:follow:user] feature in torch tensor\n",
        "edge_follow_feat_1_path = os.path.join(base_dir, \"edge-follow-feat-1.pt\")\n",
        "edge_follow_feat_1 = torch.rand(num_edges, 5)\n",
        "print(f\"Part of edge[user:follow:user] feature [feat_1]: {edge_follow_feat_1[:10, :]}\")\n",
        "torch.save(edge_follow_feat_1, edge_follow_feat_1_path)\n",
        "print(f\"Edge[user:follow:user] feature [feat_1] is saved to {edge_follow_feat_1_path}\\n\")"
      ],
      "metadata": {
        "id": "_PVu1u5brBhF"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Generate tasks\n",
        "`OnDiskDataset` supports multiple tasks. For each task, we need to prepare training/validation/test sets respectively. Such sets usually vary among different tasks. In this tutorial, let's create a **Node Classification** task and **Link Prediction** task."
      ],
      "metadata": {
        "id": "ZyqgOtsIwzh_"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Node Classification Task\n",
        "For node classification task, we need **node IDs** and corresponding **labels** for each training/validation/test set. Like feature data, numpy arrays and torch tensors are supported for these sets."
      ],
      "metadata": {
        "id": "hVxHaDIfzCkr"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# For illustration, let's generate item sets for each node type.\n",
        "num_trains = int(num_nodes * 0.6)\n",
        "num_vals = int(num_nodes * 0.2)\n",
        "num_tests = num_nodes - num_trains - num_vals\n",
        "\n",
        "user_ids = np.arange(num_nodes)\n",
        "np.random.shuffle(user_ids)\n",
        "\n",
        "item_ids = np.arange(num_nodes)\n",
        "np.random.shuffle(item_ids)\n",
        "\n",
        "# Train IDs for user.\n",
        "nc_train_user_ids_path = os.path.join(base_dir, \"nc-train-user-ids.npy\")\n",
        "nc_train_user_ids = user_ids[:num_trains]\n",
        "print(f\"Part of train ids[user] for node classification: {nc_train_user_ids[:10]}\")\n",
        "np.save(nc_train_user_ids_path, nc_train_user_ids)\n",
        "print(f\"NC train ids[user] are saved to {nc_train_user_ids_path}\\n\")\n",
        "\n",
        "# Train labels for user.\n",
        "nc_train_user_labels_path = os.path.join(base_dir, \"nc-train-user-labels.pt\")\n",
        "nc_train_user_labels = torch.randint(0, 10, (num_trains,))\n",
        "print(f\"Part of train labels[user] for node classification: {nc_train_user_labels[:10]}\")\n",
        "torch.save(nc_train_user_labels, nc_train_user_labels_path)\n",
        "print(f\"NC train labels[user] are saved to {nc_train_user_labels_path}\\n\")\n",
        "\n",
        "# Train IDs for item.\n",
        "nc_train_item_ids_path = os.path.join(base_dir, \"nc-train-item-ids.npy\")\n",
        "nc_train_item_ids = item_ids[:num_trains]\n",
        "print(f\"Part of train ids[item] for node classification: {nc_train_item_ids[:10]}\")\n",
        "np.save(nc_train_item_ids_path, nc_train_item_ids)\n",
        "print(f\"NC train ids[item] are saved to {nc_train_item_ids_path}\\n\")\n",
        "\n",
        "# Train labels for item.\n",
        "nc_train_item_labels_path = os.path.join(base_dir, \"nc-train-item-labels.pt\")\n",
        "nc_train_item_labels = torch.randint(0, 10, (num_trains,))\n",
        "print(f\"Part of train labels[item] for node classification: {nc_train_item_labels[:10]}\")\n",
        "torch.save(nc_train_item_labels, nc_train_item_labels_path)\n",
        "print(f\"NC train labels[item] are saved to {nc_train_item_labels_path}\\n\")\n",
        "\n",
        "# Val IDs for user.\n",
        "nc_val_user_ids_path = os.path.join(base_dir, \"nc-val-user-ids.npy\")\n",
        "nc_val_user_ids = user_ids[num_trains:num_trains+num_vals]\n",
        "print(f\"Part of val ids[user] for node classification: {nc_val_user_ids[:10]}\")\n",
        "np.save(nc_val_user_ids_path, nc_val_user_ids)\n",
        "print(f\"NC val ids[user] are saved to {nc_val_user_ids_path}\\n\")\n",
        "\n",
        "# Val labels for user.\n",
        "nc_val_user_labels_path = os.path.join(base_dir, \"nc-val-user-labels.pt\")\n",
        "nc_val_user_labels = torch.randint(0, 10, (num_vals,))\n",
        "print(f\"Part of val labels[user] for node classification: {nc_val_user_labels[:10]}\")\n",
        "torch.save(nc_val_user_labels, nc_val_user_labels_path)\n",
        "print(f\"NC val labels[user] are saved to {nc_val_user_labels_path}\\n\")\n",
        "\n",
        "# Val IDs for item.\n",
        "nc_val_item_ids_path = os.path.join(base_dir, \"nc-val-item-ids.npy\")\n",
        "nc_val_item_ids = item_ids[num_trains:num_trains+num_vals]\n",
        "print(f\"Part of val ids[item] for node classification: {nc_val_item_ids[:10]}\")\n",
        "np.save(nc_val_item_ids_path, nc_val_item_ids)\n",
        "print(f\"NC val ids[item] are saved to {nc_val_item_ids_path}\\n\")\n",
        "\n",
        "# Val labels for item.\n",
        "nc_val_item_labels_path = os.path.join(base_dir, \"nc-val-item-labels.pt\")\n",
        "nc_val_item_labels = torch.randint(0, 10, (num_vals,))\n",
        "print(f\"Part of val labels[item] for node classification: {nc_val_item_labels[:10]}\")\n",
        "torch.save(nc_val_item_labels, nc_val_item_labels_path)\n",
        "print(f\"NC val labels[item] are saved to {nc_val_item_labels_path}\\n\")\n",
        "\n",
        "# Test IDs for user.\n",
        "nc_test_user_ids_path = os.path.join(base_dir, \"nc-test-user-ids.npy\")\n",
        "nc_test_user_ids = user_ids[-num_tests:]\n",
        "print(f\"Part of test ids[user] for node classification: {nc_test_user_ids[:10]}\")\n",
        "np.save(nc_test_user_ids_path, nc_test_user_ids)\n",
        "print(f\"NC test ids[user] are saved to {nc_test_user_ids_path}\\n\")\n",
        "\n",
        "# Test labels for user.\n",
        "nc_test_user_labels_path = os.path.join(base_dir, \"nc-test-user-labels.pt\")\n",
        "nc_test_user_labels = torch.randint(0, 10, (num_tests,))\n",
        "print(f\"Part of test labels[user] for node classification: {nc_test_user_labels[:10]}\")\n",
        "torch.save(nc_test_user_labels, nc_test_user_labels_path)\n",
        "print(f\"NC test labels[user] are saved to {nc_test_user_labels_path}\\n\")\n",
        "\n",
        "# Test IDs for item.\n",
        "nc_test_item_ids_path = os.path.join(base_dir, \"nc-test-item-ids.npy\")\n",
        "nc_test_item_ids = item_ids[-num_tests:]\n",
        "print(f\"Part of test ids[item] for node classification: {nc_test_item_ids[:10]}\")\n",
        "np.save(nc_test_item_ids_path, nc_test_item_ids)\n",
        "print(f\"NC test ids[item] are saved to {nc_test_item_ids_path}\\n\")\n",
        "\n",
        "# Test labels for item.\n",
        "nc_test_item_labels_path = os.path.join(base_dir, \"nc-test-item-labels.pt\")\n",
        "nc_test_item_labels = torch.randint(0, 10, (num_tests,))\n",
        "print(f\"Part of test labels[item] for node classification: {nc_test_item_labels[:10]}\")\n",
        "torch.save(nc_test_item_labels, nc_test_item_labels_path)\n",
        "print(f\"NC test labels[item] are saved to {nc_test_item_labels_path}\\n\")"
      ],
      "metadata": {
        "id": "S5-fyBbHzTCO"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Link Prediction Task\n",
        "For link prediction task, we need **node pairs** or **negative src/dsts** for each training/validation/test set. Like feature data, numpy arrays and torch tensors are supported for these sets."
      ],
      "metadata": {
        "id": "LhAcDCHQ_KJ0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# For illustration, let's generate item sets for each edge type.\n",
        "num_trains = int(num_edges * 0.6)\n",
        "num_vals = int(num_edges * 0.2)\n",
        "num_tests = num_edges - num_trains - num_vals\n",
        "\n",
        "# Train node pairs for user:like:item.\n",
        "lp_train_like_node_pairs_path = os.path.join(base_dir, \"lp-train-like-node-pairs.npy\")\n",
        "lp_train_like_node_pairs = like_edges[:num_trains, :]\n",
        "print(f\"Part of train node pairs[user:like:item] for link prediction: {lp_train_like_node_pairs[:10]}\")\n",
        "np.save(lp_train_like_node_pairs_path, lp_train_like_node_pairs)\n",
        "print(f\"LP train node pairs[user:like:item] are saved to {lp_train_like_node_pairs_path}\\n\")\n",
        "\n",
        "# Train node pairs for user:follow:user.\n",
        "lp_train_follow_node_pairs_path = os.path.join(base_dir, \"lp-train-follow-node-pairs.npy\")\n",
        "lp_train_follow_node_pairs = follow_edges[:num_trains, :]\n",
        "print(f\"Part of train node pairs[user:follow:user] for link prediction: {lp_train_follow_node_pairs[:10]}\")\n",
        "np.save(lp_train_follow_node_pairs_path, lp_train_follow_node_pairs)\n",
        "print(f\"LP train node pairs[user:follow:user] are saved to {lp_train_follow_node_pairs_path}\\n\")\n",
        "\n",
        "# Val node pairs for user:like:item.\n",
        "lp_val_like_node_pairs_path = os.path.join(base_dir, \"lp-val-like-node-pairs.npy\")\n",
        "lp_val_like_node_pairs = like_edges[num_trains:num_trains+num_vals, :]\n",
        "print(f\"Part of val node pairs[user:like:item] for link prediction: {lp_val_like_node_pairs[:10]}\")\n",
        "np.save(lp_val_like_node_pairs_path, lp_val_like_node_pairs)\n",
        "print(f\"LP val node pairs[user:like:item] are saved to {lp_val_like_node_pairs_path}\\n\")\n",
        "\n",
        "# Val negative dsts for user:like:item.\n",
        "lp_val_like_neg_dsts_path = os.path.join(base_dir, \"lp-val-like-neg-dsts.pt\")\n",
        "lp_val_like_neg_dsts = torch.randint(0, num_nodes, (num_vals, 10))\n",
        "print(f\"Part of val negative dsts[user:like:item] for link prediction: {lp_val_like_neg_dsts[:10]}\")\n",
        "torch.save(lp_val_like_neg_dsts, lp_val_like_neg_dsts_path)\n",
        "print(f\"LP val negative dsts[user:like:item] are saved to {lp_val_like_neg_dsts_path}\\n\")\n",
        "\n",
        "# Val node pairs for user:follow:user.\n",
        "lp_val_follow_node_pairs_path = os.path.join(base_dir, \"lp-val-follow-node-pairs.npy\")\n",
        "lp_val_follow_node_pairs = follow_edges[num_trains:num_trains+num_vals, :]\n",
        "print(f\"Part of val node pairs[user:follow:user] for link prediction: {lp_val_follow_node_pairs[:10]}\")\n",
        "np.save(lp_val_follow_node_pairs_path, lp_val_follow_node_pairs)\n",
        "print(f\"LP val node pairs[user:follow:user] are saved to {lp_val_follow_node_pairs_path}\\n\")\n",
        "\n",
        "# Val negative dsts for user:follow:user.\n",
        "lp_val_follow_neg_dsts_path = os.path.join(base_dir, \"lp-val-follow-neg-dsts.pt\")\n",
        "lp_val_follow_neg_dsts = torch.randint(0, num_nodes, (num_vals, 10))\n",
        "print(f\"Part of val negative dsts[user:follow:user] for link prediction: {lp_val_follow_neg_dsts[:10]}\")\n",
        "torch.save(lp_val_follow_neg_dsts, lp_val_follow_neg_dsts_path)\n",
        "print(f\"LP val negative dsts[user:follow:user] are saved to {lp_val_follow_neg_dsts_path}\\n\")\n",
        "\n",
        "# Test node paris for user:like:item.\n",
        "lp_test_like_node_pairs_path = os.path.join(base_dir, \"lp-test-like-node-pairs.npy\")\n",
        "lp_test_like_node_pairs = like_edges[-num_tests, :]\n",
        "print(f\"Part of test node pairs[user:like:item] for link prediction: {lp_test_like_node_pairs[:10]}\")\n",
        "np.save(lp_test_like_node_pairs_path, lp_test_like_node_pairs)\n",
        "print(f\"LP test node pairs[user:like:item] are saved to {lp_test_like_node_pairs_path}\\n\")\n",
        "\n",
        "# Test negative dsts for user:like:item.\n",
        "lp_test_like_neg_dsts_path = os.path.join(base_dir, \"lp-test-like-neg-dsts.pt\")\n",
        "lp_test_like_neg_dsts = torch.randint(0, num_nodes, (num_tests, 10))\n",
        "print(f\"Part of test negative dsts[user:like:item] for link prediction: {lp_test_like_neg_dsts[:10]}\")\n",
        "torch.save(lp_test_like_neg_dsts, lp_test_like_neg_dsts_path)\n",
        "print(f\"LP test negative dsts[user:like:item] are saved to {lp_test_like_neg_dsts_path}\\n\")\n",
        "\n",
        "# Test node paris for user:follow:user.\n",
        "lp_test_follow_node_pairs_path = os.path.join(base_dir, \"lp-test-follow-node-pairs.npy\")\n",
        "lp_test_follow_node_pairs = follow_edges[-num_tests, :]\n",
        "print(f\"Part of test node pairs[user:follow:user] for link prediction: {lp_test_follow_node_pairs[:10]}\")\n",
        "np.save(lp_test_follow_node_pairs_path, lp_test_follow_node_pairs)\n",
        "print(f\"LP test node pairs[user:follow:user] are saved to {lp_test_follow_node_pairs_path}\\n\")\n",
        "\n",
        "# Test negative dsts for user:follow:user.\n",
        "lp_test_follow_neg_dsts_path = os.path.join(base_dir, \"lp-test-follow-neg-dsts.pt\")\n",
        "lp_test_follow_neg_dsts = torch.randint(0, num_nodes, (num_tests, 10))\n",
        "print(f\"Part of test negative dsts[user:follow:user] for link prediction: {lp_test_follow_neg_dsts[:10]}\")\n",
        "torch.save(lp_test_follow_neg_dsts, lp_test_follow_neg_dsts_path)\n",
        "print(f\"LP test negative dsts[user:follow:user] are saved to {lp_test_follow_neg_dsts_path}\\n\")"
      ],
      "metadata": {
        "id": "u0jCnXIcAQy4"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Organize Data into YAML File\n",
        "Now we need to create a `metadata.yaml` file which contains the paths, dadta types of graph structure, feature data, training/validation/test sets. Please note that all path should be relative to `metadata.yaml`.\n",
        "\n",
        "For heterogeneous graph, we need to specify the node/edge type in **type** fields. For edge type, canonical etype is required which is a string that's concatenated by source node type, etype, and destination node type together with `:`.\n",
        "\n",
        "Notes:\n",
        "- all path should be relative to `metadata.yaml`.\n",
        "- Below fields are optional and not specified in below example.\n",
        "  - `in_memory`: indicates whether to load dada into memory or `mmap`. Default is `True`.\n",
        "\n",
        "Please refer to [YAML specification](https://github.com/dmlc/dgl/blob/master/docs/source/stochastic_training/ondisk-dataset-specification.rst) for more details."
      ],
      "metadata": {
        "id": "wbk6-wxRK-6S"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "yaml_content = f\"\"\"\n",
        "    dataset_name: heterogeneous_graph_nc_lp\n",
        "    graph:\n",
        "      nodes:\n",
        "        - type: user\n",
        "          num: {num_nodes}\n",
        "        - type: item\n",
        "          num: {num_nodes}\n",
        "      edges:\n",
        "        - type: \"user:like:item\"\n",
        "          format: csv\n",
        "          path: {os.path.basename(like_edges_path)}\n",
        "        - type: \"user:follow:user\"\n",
        "          format: csv\n",
        "          path: {os.path.basename(follow_edges_path)}\n",
        "    feature_data:\n",
        "      - domain: node\n",
        "        type: user\n",
        "        name: feat_0\n",
        "        format: numpy\n",
        "        path: {os.path.basename(node_user_feat_0_path)}\n",
        "      - domain: node\n",
        "        type: user\n",
        "        name: feat_1\n",
        "        format: torch\n",
        "        path: {os.path.basename(node_user_feat_1_path)}\n",
        "      - domain: node\n",
        "        type: item\n",
        "        name: feat_0\n",
        "        format: numpy\n",
        "        path: {os.path.basename(node_item_feat_0_path)}\n",
        "      - domain: node\n",
        "        type: item\n",
        "        name: feat_1\n",
        "        format: torch\n",
        "        path: {os.path.basename(node_item_feat_1_path)}\n",
        "      - domain: edge\n",
        "        type: \"user:like:item\"\n",
        "        name: feat_0\n",
        "        format: numpy\n",
        "        path: {os.path.basename(edge_like_feat_0_path)}\n",
        "      - domain: edge\n",
        "        type: \"user:like:item\"\n",
        "        name: feat_1\n",
        "        format: torch\n",
        "        path: {os.path.basename(edge_like_feat_1_path)}\n",
        "      - domain: edge\n",
        "        type: \"user:follow:user\"\n",
        "        name: feat_0\n",
        "        format: numpy\n",
        "        path: {os.path.basename(edge_follow_feat_0_path)}\n",
        "      - domain: edge\n",
        "        type: \"user:follow:user\"\n",
        "        name: feat_1\n",
        "        format: torch\n",
        "        path: {os.path.basename(edge_follow_feat_1_path)}\n",
        "    tasks:\n",
        "      - name: node_classification\n",
        "        num_classes: 10\n",
        "        train_set:\n",
        "          - data:\n",
        "              - type: user\n",
        "                name: seed_nodes\n",
        "                format: numpy\n",
        "                path: {os.path.basename(nc_train_user_ids_path)}\n",
        "              - type: user\n",
        "                name: labels\n",
        "                format: torch\n",
        "                path: {os.path.basename(nc_train_user_labels_path)}\n",
        "              - type: item\n",
        "                name: seed_nodes\n",
        "                format: numpy\n",
        "                path: {os.path.basename(nc_train_item_ids_path)}\n",
        "              - type: item\n",
        "                name: labels\n",
        "                format: torch\n",
        "                path: {os.path.basename(nc_train_item_labels_path)}\n",
        "        validation_set:\n",
        "          - data:\n",
        "              - type: user\n",
        "                name: seed_nodes\n",
        "                format: numpy\n",
        "                path: {os.path.basename(nc_val_user_ids_path)}\n",
        "              - type: user\n",
        "                name: labels\n",
        "                format: torch\n",
        "                path: {os.path.basename(nc_val_user_labels_path)}\n",
        "              - type: item\n",
        "                name: seed_nodes\n",
        "                format: numpy\n",
        "                path: {os.path.basename(nc_val_item_ids_path)}\n",
        "              - type: item\n",
        "                name: labels\n",
        "                format: torch\n",
        "                path: {os.path.basename(nc_val_item_labels_path)}\n",
        "        test_set:\n",
        "          - data:\n",
        "              - type: user\n",
        "                name: seed_nodes\n",
        "                format: numpy\n",
        "                path: {os.path.basename(nc_test_user_ids_path)}\n",
        "              - type: user\n",
        "                name: labels\n",
        "                format: torch\n",
        "                path: {os.path.basename(nc_test_user_labels_path)}\n",
        "              - type: item\n",
        "                name: seed_nodes\n",
        "                format: numpy\n",
        "                path: {os.path.basename(nc_test_item_ids_path)}\n",
        "              - type: item\n",
        "                name: labels\n",
        "                format: torch\n",
        "                path: {os.path.basename(nc_test_item_labels_path)}\n",
        "      - name: link_prediction\n",
        "        num_classes: 10\n",
        "        train_set:\n",
        "          - data:\n",
        "              - type: \"user:like:item\"\n",
        "                name: node_pairs\n",
        "                format: numpy\n",
        "                path: {os.path.basename(lp_train_like_node_pairs_path)}\n",
        "              - type: \"user:follow:user\"\n",
        "                name: node_pairs\n",
        "                format: numpy\n",
        "                path: {os.path.basename(lp_train_follow_node_pairs_path)}\n",
        "        validation_set:\n",
        "          - data:\n",
        "              - type: \"user:like:item\"\n",
        "                name: node_pairs\n",
        "                format: numpy\n",
        "                path: {os.path.basename(lp_val_like_node_pairs_path)}\n",
        "              - type: \"user:like:item\"\n",
        "                name: negative_dsts\n",
        "                format: torch\n",
        "                path: {os.path.basename(lp_val_like_neg_dsts_path)}\n",
        "              - type: \"user:follow:user\"\n",
        "                name: node_pairs\n",
        "                format: numpy\n",
        "                path: {os.path.basename(lp_val_follow_node_pairs_path)}\n",
        "              - type: \"user:follow:user\"\n",
        "                name: negative_dsts\n",
        "                format: torch\n",
        "                path: {os.path.basename(lp_val_follow_neg_dsts_path)}\n",
        "        test_set:\n",
        "          - data:\n",
        "              - type: \"user:like:item\"\n",
        "                name: node_pairs\n",
        "                format: numpy\n",
        "                path: {os.path.basename(lp_test_like_node_pairs_path)}\n",
        "              - type: \"user:like:item\"\n",
        "                name: negative_dsts\n",
        "                format: torch\n",
        "                path: {os.path.basename(lp_test_like_neg_dsts_path)}\n",
        "              - type: \"user:follow:user\"\n",
        "                name: node_pairs\n",
        "                format: numpy\n",
        "                path: {os.path.basename(lp_test_follow_node_pairs_path)}\n",
        "              - type: \"user:follow:user\"\n",
        "                name: negative_dsts\n",
        "                format: torch\n",
        "                path: {os.path.basename(lp_test_follow_neg_dsts_path)}\n",
        "\"\"\"\n",
        "metadata_path = os.path.join(base_dir, \"metadata.yaml\")\n",
        "with open(metadata_path, \"w\") as f:\n",
        "  f.write(yaml_content)"
      ],
      "metadata": {
        "id": "ddGTWW61Lpwp"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Instantiate `OnDiskDataset`\n",
        "Now we're ready to load dataset via `dgl.graphbolt.OnDiskDataset`. When instantiating, we just pass in the base directory where `metadata.yaml` file lies.\n",
        "\n",
        "During first instantiation, GraphBolt preprocesses the raw data such as constructing `FusedCSCSamplingGraph` from edges. All data including graph, feature data, training/validation/test sets are put into `preprocessed` directory after preprocessing. Any following dataset loading will skip the preprocess stage.\n",
        "\n",
        "After preprocessing, `load()` is required to be called explicitly in order to load graph, feature data and tasks."
      ],
      "metadata": {
        "id": "kEfybHGhOW7O"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "dataset = gb.OnDiskDataset(base_dir).load()\n",
        "graph = dataset.graph\n",
        "print(f\"Loaded graph: {graph}\\n\")\n",
        "\n",
        "feature = dataset.feature\n",
        "print(f\"Loaded feature store: {feature}\\n\")\n",
        "\n",
        "tasks = dataset.tasks\n",
        "nc_task = tasks[0]\n",
        "print(f\"Loaded node classification task: {nc_task}\\n\")\n",
        "lp_task = tasks[1]\n",
        "print(f\"Loaded link prediction task: {lp_task}\\n\")"
      ],
      "metadata": {
        "id": "W58CZoSzOiyo"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}