[BugFix] fix problems in data split (#3082)

* [BugFix] fix problems in data split

* fix format problems in docstring

* modify statistics to fit in dgl nature
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>
Co-authored-by: zhjwy9343 <6593865@qq.com>

python/dgl/data/fakenews.py

@@ -19,14 +19,18 @@ class FakeNewsDataset(DGLBuiltinDataset):
     the root node represents the news, the leaf nodes are Twitter users
     who retweeted the root news. Besides, the node features are encoded
     user historical tweets using different pretrained language models:
-        bert: the 768-dimensional node feature composed of Twitter user
+
+    - bert: the 768-dimensional node feature composed of Twitter user
     historical tweets encoded by the bert-as-service
-        content: the 310-dimensional node feature composed of a
+
+    - content: the 310-dimensional node feature composed of a
     300-dimensional “spacy” vector plus a 10-dimensional
     “profile” vector
-        profile: the 10-dimensional node feature composed of ten Twitter
+
+    - profile: the 10-dimensional node feature composed of ten Twitter
     user profile attributes.
-        spacy: the 300-dimensional node feature composed of Twitter user
+
+    - spacy: the 300-dimensional node feature composed of Twitter user
     historical tweets encoded by the spaCy word2vec encoder.
 
     Note: this dataset is for academic use only, and commercial use is prohibited.
@@ -39,27 +43,33 @@ class FakeNewsDataset(DGLBuiltinDataset):
         - Nodes: 41,054
         - Edges: 40,740
         - Classes:
-            Fake: 157
-            Real: 157
+
+            - Fake: 157
+            - Real: 157
+
         - Node feature size:
-            bert: 768
-            content: 310
-            profile: 10
-            spacy: 300
+
+            - bert: 768
+            - content: 310
+            - profile: 10
+            - spacy: 300
 
         Gossipcop:
 
-        - Graphs: 5464
+        - Graphs: 5,464
         - Nodes: 314,262
         - Edges: 308,798
         - Classes:
-            Fake: 2732
-            Real: 2732
+
+            - Fake: 2,732
+            - Real: 2,732
+
         - Node feature size:
-            bert: 768
-            content: 310
-            profile: 10
-            spacy: 300
+
+            - bert: 768
+            - content: 310
+            - profile: 10
+            - spacy: 300
 
     Parameters
     ----------

python/dgl/data/fraud.py

@@ -177,15 +177,15 @@ class FraudDataset(DGLBuiltinDataset):
             "must between 0 and 1 (inclusive)."
 
         N = x.shape[0]
-        index = list(range(N))
+        index = np.arange(N)
         if self.name == 'amazon':
             # 0-3304 are unlabeled nodes
-            index = list(range(3305, N))
+            index = np.arange(3305, N)
 
-        np.random.RandomState(seed).permutation(index)
-        train_idx = index[:int(train_size * N)]
-        val_idx = index[int(N - val_size * N):]
-        test_idx = index[int(train_size * N):int(N - val_size * N)]
+        index = np.random.RandomState(seed).permutation(index)
+        train_idx = index[:int(train_size * len(index))]
+        val_idx = index[len(index) - int(val_size * len(index)):]
+        test_idx = index[int(train_size * len(index)):len(index) - int(val_size * len(index))]
         train_mask = np.zeros(N, dtype=np.bool)
         val_mask = np.zeros(N, dtype=np.bool)
         test_mask = np.zeros(N, dtype=np.bool)
@@ -202,9 +202,9 @@ class FraudYelpDataset(FraudDataset):
 
     The Yelp dataset includes hotel and restaurant reviews filtered (spam) and recommended
     (legitimate) by Yelp. A spam review detection task can be conducted, which is a binary
-    classification task. 32 handcrafted features from
-    <http://dx.doi.org/10.1145/2783258.2783370> are taken as the raw node features. Reviews
-    are nodes in the graph, and three relations are:
+    classification task. 32 handcrafted features from <http://dx.doi.org/10.1145/2783258.2783370>
+    are taken as the raw node features. Reviews are nodes in the graph, and three relations are:
+
         1. R-U-R: it connects reviews posted by the same user
         2. R-S-R: it connects reviews under the same product with the same star rating (1-5 stars)
         3. R-T-R: it connects two reviews under the same product posted in the same month.
@@ -213,13 +213,16 @@ class FraudYelpDataset(FraudDataset):
 
     - Nodes: 45,954
     - Edges:
-        R-U-R: 49,315
-        R-T-R: 573,616
-        R-S-R: 3,402,743
-        ALL: 3,846,979
+
+        - R-U-R: 98,630
+        - R-T-R: 1,147,232
+        - R-S-R: 6,805,486
+
     - Classes:
-        Positive (spam): 6,677
-        Negative (legitimate): 39,277
+
+        - Positive (spam): 6,677
+        - Negative (legitimate): 39,277
+
     - Positive-Negative ratio: 1 : 5.9
     - Node feature size: 32
 
@@ -278,14 +281,18 @@ class FraudAmazonDataset(FraudDataset):
 
     - Nodes: 11,944
     - Edges:
-        U-P-U: 175,608
-        U-S-U: 3,566,479
-        U-V-U: 1,036,737
-        ALL: 4,398,392
+
+        - U-P-U: 351,216
+        - U-S-U: 7,132,958
+        - U-V-U: 2,073,474
+
     - Classes:
-        Positive (fraudulent): 821
-        Negative (benign): 11,123
-    - Positive-Negative ratio: 1 : 13.5
+
+        - Positive (fraudulent): 821
+        - Negative (benign): 7,818
+        - Unlabeled: 3,305
+
+    - Positive-Negative ratio: 1 : 10.5
     - Node feature size: 25
 
     Parameters