Upload 2 files

Amazon-APP/processed/Amazon-Movies-processing.py

@@ -0,0 +1,136 @@
+import json
+from tqdm import tqdm
+
+def read_json_lines(file_path):
+    data = []
+    with open(file_path, 'r') as file:
+        for line in file:
+            try:
+                data.append(json.loads(line.strip()))
+            except json.JSONDecodeError as e:
+                print(f"Error decoding JSON on line: {line}")
+                print(e)
+    return data
+
+# Define file paths
+reviews_path = '/home/ubuntu/tmp/pycharm_project_714/GNN/Movies_and_TV_5.json'
+meta_path = '/home/ubuntu/tmp/pycharm_project_714/GNN/meta_Movies_and_TV.json'
+
+# Read and preview JSON files
+reviews_data = read_json_lines(reviews_path)
+meta_data = read_json_lines(meta_path)
+
+
+reviewer_nodes = []
+asin_nodes = []
+reviewer_id2idx = {}
+asin_id2idx = {}
+
+
+for review in tqdm(reviews_data, desc="Processing reviews"):
+    reviewer_id = review.get('reviewerID')
+    if reviewer_id and reviewer_id not in reviewer_id2idx:
+        reviewer_id2idx[reviewer_id] = len(reviewer_nodes)
+        reviewer_nodes.append({reviewer_id: reviewer_id2idx[reviewer_id]})
+
+for review in tqdm(reviews_data, desc="Processing reviews"):
+    asin_id = review.get('asin')
+    if asin_id and asin_id not in asin_id2idx:
+        asin_id2idx[asin_id] = len(asin_nodes)
+        asin_nodes.append({asin_id: len(reviewer_nodes) + asin_id2idx[asin_id]})
+
+
+print(f"Total unique reviewers: {len(reviewer_nodes)}")
+print(f"Total unique ASINs: {len(asin_nodes)}")
+
+
+print("Sample reviewer nodes:", reviewer_nodes[:5])
+print("Sample asin nodes:", asin_nodes[:5])
+
+
+reviewer_nodes_text = [""] * len(reviewer_nodes)
+reviewer_nodes_label = ["Unknown"] * len(reviewer_nodes)
+asin_nodes_text = [""] * len(asin_nodes)
+asin_nodes_label = ["Unknown"] * len(asin_nodes)
+added_asin_summaries = set()
+
+for meta in tqdm(meta_data, desc="Processing meta data"):
+    asin_id = meta.get('asin')
+    categories = meta.get('category')
+    if asin_id in asin_id2idx:
+        asin_idx = asin_id2idx[asin_id]
+        if len(categories) > 1:
+            asin_nodes_label[asin_idx] = categories[1]
+        else:
+            asin_nodes_label[asin_idx] = categories[0]
+
+
+for review in tqdm(reviews_data, desc="Adding summary to ASIN nodes"):
+    asin_id = review.get('asin')
+    summary = review.get('summary', 'No summary available.')
+
+    if asin_id in asin_id2idx and asin_id not in added_asin_summaries:
+        asin_idx = asin_id2idx[asin_id]
+        asin_nodes_text[asin_idx] = summary
+        added_asin_summaries.add(asin_id)
+
+print("Sample ASIN nodes text:", asin_nodes_text[:5])
+print("Sample ASIN nodes label:", asin_nodes_label[:5])
+
+edge_index = [[], []]
+edge_attr = []
+
+# 解析评论数据，生成 edge_index 和 edge_attr
+for review in tqdm(reviews_data, desc="Creating edges"):
+    reviewer_id = review['reviewerID']
+    asin_id = review['asin']
+    review_text = review['reviewText']
+
+    if reviewer_id in reviewer_id2idx and asin_id in asin_id2idx:
+        reviewer_idx = reviewer_id2idx[reviewer_id]
+        asin_idx = asin_id2idx[asin_id] + len(reviewer_nodes)
+
+        edge_index[0].append(reviewer_idx)
+        edge_index[1].append(asin_idx)
+        edge_attr.append(review_text)
+
+
+print(f"Total edges: {len(edge_index[0])}")
+print("Sample edges:", list(zip(edge_index[0][:5], edge_index[1][:5], edge_attr[:5])))
+for meta in tqdm(meta_data, desc="Processing also_bought edges"):
+    asin_id = meta['asin']
+    also_bought = meta.get('related', {}).get('also_bought', [])
+
+    if asin_id in asin_id2idx:
+        asin_idx = asin_id2idx[asin_id]
+        for related_asin in also_bought:
+            if related_asin not in asin_id2idx:
+                asin_id2idx[related_asin] = len(asin_nodes) + len(reviewer_nodes)
+                asin_nodes.append({related_asin: asin_id2idx[related_asin]})
+                asin_nodes_text.append("")
+                asin_nodes_label.append("Unknown")
+
+            related_asin_idx = asin_id2idx[related_asin]
+
+            edge_index[0].append(asin_idx)
+            edge_index[1].append(related_asin_idx)
+            edge_attr.append("")
+
+print(f"Total edges after adding also_bought: {len(edge_index[0])}")
+print("Sample edges:", list(zip(edge_index[0][-5:], edge_index[1][-5:], edge_attr[-5:])))
+#%%
+all_nodes = reviewer_nodes + asin_nodes
+text_nodes = ["reviewer"] * len(reviewer_nodes) + ["asin"] * len(asin_nodes)
+#%%
+nodes_text = reviewer_nodes_text + asin_nodes_text
+nodes_text_labels = reviewer_nodes_label + asin_nodes_label
+#%%
+import torch
+from torch_geometric.data import Data
+graph = Data(
+    text_nodes=text_nodes,
+    edge_index=torch.tensor(edge_index, dtype=torch.long),
+    text_edges=edge_attr,
+    node_texts=nodes_text,
+    text_node_labels=nodes_text_labels
+)

Amazon-APP/processed/download_data.sh

@@ -0,0 +1,24 @@
+#!/bin/bash
+
+SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
+BASE_DIR="$(dirname "$SCRIPT_DIR")"
+RAW_DIR="$BASE_DIR/processed/original"
+
+mkdir -p "$RAW_DIR"
+
+urls=(
+    "https://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Apps_for_Android_5.json.gz"
+    "https://snap.stanford.edu/data/amazon/productGraph/categoryFiles/meta_Apps_for_Android.json.gz"
+)
+
+for url in "${urls[@]}"; do
+    wget -P "$RAW_DIR" "$url"
+done
+
+for file in "$RAW_DIR"/*.gz; do
+    gunzip "$file"
+done
+
+echo "Download and extraction complete."
+
+