diff --git a/recommend/inference.py b/recommend/inference.py
index 697b569..346209c 100644
--- a/recommend/inference.py
+++ b/recommend/inference.py
@@ -3,7 +3,7 @@
 
 from os import path
 from utils.parse_args import args
-from utils.dataloader import EdgeListData
+from utils.data_loader import EdgeListData
 from model.LightGCN import LightGCN
 import torch
 import numpy as np
@@ -13,15 +13,14 @@
 t_start = time.time()
 
 # 配置参数
-args.data_path = './'
 args.device = 'cuda:7'
+args.data_path = './user_seed_graph.txt'
 args.pre_model_path = './model/LightGCN_pretrained.pt'
 
+
 # 1. 加载数据集
 t_data_start = time.time()
-pretrain_data = path.join(args.data_path, "uig.txt")
-pretrain_val_data = path.join(args.data_path, "uig.txt")
-dataset = EdgeListData(pretrain_data, pretrain_val_data)
+dataset = EdgeListData(args.data_path, args.data_path)
 t_data_end = time.time()
 
 
diff --git a/recommend/uig.txt b/recommend/uig.txt
deleted file mode 100644
index 5846057..0000000
--- a/recommend/uig.txt
+++ /dev/null
@@ -1,2 +0,0 @@
-0	1 3 9 12 5 7 6 8 4	1511683379 1511683385 1511683431 1511683453 1511683481 1511692992 1511693011 1511693077 1511787191
-1	10 11 2	1511578239 1511594732 1511664627
\ No newline at end of file
diff --git a/recommend/user_seed_graph.txt b/recommend/user_seed_graph.txt
new file mode 100644
index 0000000..ef29920
--- /dev/null
+++ b/recommend/user_seed_graph.txt
@@ -0,0 +1,3 @@
+0	0 0 1 2	1746061954 1736237924 1736240066 1736309966
+1	1 1	1746315010 1746583706
+2	2 0 2	1746738305 1746865166 1749284366
diff --git a/recommend/utils/data_generator.py b/recommend/utils/data_generator.py
new file mode 100644
index 0000000..e50b4e9
--- /dev/null
+++ b/recommend/utils/data_generator.py
@@ -0,0 +1,87 @@
+import pymysql
+import datetime
+from collections import defaultdict
+
+SqlURL = "10.126.59.25"
+SqlPort = 3306
+Database = "pt_database_test"
+SqlUsername = "root"
+SqlPassword = "123456"
+
+
+def fetch_data():
+    conn = pymysql.connect(
+        host=SqlURL,
+        port=SqlPort,
+        user=SqlUsername,
+        password=SqlPassword,
+        database=Database,
+        charset="utf8mb4"
+    )
+    cursor = conn.cursor()
+    cursor.execute("SELECT user_id, seed_id, download_start FROM SeedDownload")
+    download_rows = cursor.fetchall()
+    cursor.execute("SELECT user_id, seed_id, created_at FROM UserFavorite")
+    favorite_rows = cursor.fetchall()
+    cursor.close()
+    conn.close()
+    return download_rows, favorite_rows
+
+
+def process_records(download_rows, favorite_rows):
+    records = []
+    user_set = set()
+    seed_set = set()
+    for row in download_rows:
+        user_id, seed_id, created_at = row
+        user_set.add(user_id)
+        seed_set.add(seed_id)
+        if isinstance(created_at, datetime.datetime):
+            ts = int(created_at.timestamp())
+        else:
+            ts = 0
+        records.append((user_id, seed_id, ts))
+    for row in favorite_rows:
+        user_id, seed_id, created_at = row
+        user_set.add(user_id)
+        seed_set.add(seed_id)
+        if isinstance(created_at, datetime.datetime):
+            ts = int(created_at.timestamp())
+        else:
+            ts = 0
+        records.append((user_id, seed_id, ts))
+    return records, user_set, seed_set
+
+
+def build_id_maps(user_set, seed_set):
+    user2idx = {uid: idx for idx, uid in enumerate(sorted(user_set))}
+    seed2idx = {sid: idx for idx, sid in enumerate(sorted(seed_set))}
+    return user2idx, seed2idx
+
+
+def group_and_write(records, user2idx, seed2idx, output_path="./user_seed_graph.txt"):
+    user_items = defaultdict(list)
+    user_times = defaultdict(list)
+    for user_id, seed_id, ts in records:
+        uid = user2idx[user_id]
+        sid = seed2idx[seed_id]
+        user_items[uid].append(sid)
+        user_times[uid].append(ts)
+    print(user_items)
+    print(user_times)
+    with open(output_path, "w", encoding="utf-8") as f:
+        for uid in sorted(user_items.keys()):
+            items = " ".join(str(item) for item in user_items[uid])
+            times = " ".join(str(t) for t in user_times[uid])
+            f.write(f"{uid}\t{items}\t{times}\n")
+
+
+def main():
+    download_rows, favorite_rows = fetch_data()
+    records, user_set, seed_set = process_records(download_rows, favorite_rows)
+    user2idx, seed2idx = build_id_maps(user_set, seed_set)
+    group_and_write(records, user2idx, seed2idx)
+
+
+if __name__ == "__main__":
+    main()
\ No newline at end of file
diff --git a/recommend/utils/dataloader.py b/recommend/utils/data_loader.py
similarity index 100%
rename from recommend/utils/dataloader.py
rename to recommend/utils/data_loader.py