[BugFix] fix problems found in bug bash (#3116)

* fix breakline in fakenews.py

* fix inconsistent argument name

* modify incorrect example and deprecated graph type

* modify docstring and example in knn_graph

* fix incorrect node type
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>

examples/pytorch/graphsim/README.md

@@ -43,17 +43,17 @@ python n_body_sim.py --num_traj <num_traj> --steps <num_steps>
 
 
 ```python
-python train.py --number_workers 15
+python train.py --num_workers 15
 ```
 
 Training with GPU
 ```python
-python train.py --gpu 0 --number_workers 15
+python train.py --gpu 0 --num_workers 15
 ```
 
 Training with visualization: for valid visualization, it might take full 40000 epoch of training
 ```python
-python train.py --gpu 0 --number_workers 15 --visualize
+python train.py --gpu 0 --num_workers 15 --visualize
 ```
 
 One Step Loss Performance, Loss of test data after 40000 training epochs.

python/dgl/data/fakenews.py

@@ -20,18 +20,10 @@ class FakeNewsDataset(DGLBuiltinDataset):
     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
-    historical tweets encoded by the bert-as-service
-
-    - 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
-    user profile attributes.
-
-    - spacy: the 300-dimensional node feature composed of Twitter user
-    historical tweets encoded by the spaCy word2vec encoder.
+    - 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 300-dimensional “spacy” vector plus a 10-dimensional “profile” vector
+    - profile: the 10-dimensional node feature composed of ten Twitter user profile attributes.
+    - 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.
 

python/dgl/data/fraud.py

@@ -51,7 +51,7 @@ class FraudDataset(DGLBuiltinDataset):
     ----------
     num_classes : int
         Number of label classes
-    graph : dgl.heterograph.DGLHeteroGraph
+    graph : dgl.DGLGraph
         Graph structure, etc.
     seed : int
         Random seed in splitting the dataset.
@@ -65,8 +65,8 @@ class FraudDataset(DGLBuiltinDataset):
     >>> dataset = FraudDataset('yelp')
     >>> graph = dataset[0]
     >>> num_classes = dataset.num_classes
-    >>> feat = dataset.ndata['feature']
-    >>> label = dataset.ndata['label']
+    >>> feat = graph.ndata['feature']
+    >>> label = graph.ndata['label']
     """
     file_urls = {
         'yelp': 'dataset/FraudYelp.zip',
@@ -81,8 +81,8 @@ class FraudDataset(DGLBuiltinDataset):
         'amazon': 'Amazon.mat'
     }
     node_name = {
-        'yelp': 'user',
-        'amazon': 'review'
+        'yelp': 'review',
+        'amazon': 'user'
     }
 
     def __init__(self, name, raw_dir=None, random_seed=717, train_size=0.7, val_size=0.1):
@@ -126,7 +126,7 @@ class FraudDataset(DGLBuiltinDataset):
 
         Returns
         -------
-        :class:`dgl.heterograph.DGLHeteroGraph`
+        :class:`dgl.DGLGraph`
             graph structure, node features, node labels and masks
 
             - ``ndata['feature']``: node features
@@ -249,8 +249,8 @@ class FraudYelpDataset(FraudDataset):
     >>> dataset = FraudYelpDataset()
     >>> graph = dataset[0]
     >>> num_classes = dataset.num_classes
-    >>> feat = dataset.ndata['feature']
-    >>> label = dataset.ndata['label']
+    >>> feat = graph.ndata['feature']
+    >>> label = graph.ndata['label']
     """
 
     def __init__(self, raw_dir=None, random_seed=717, train_size=0.7, val_size=0.1):
@@ -318,8 +318,8 @@ class FraudAmazonDataset(FraudDataset):
     >>> dataset = FraudAmazonDataset()
     >>> graph = dataset[0]
     >>> num_classes = dataset.num_classes
-    >>> feat = dataset.ndata['feature']
-    >>> label = dataset.ndata['label']
+    >>> feat = graph.ndata['feature']
+    >>> label = graph.ndata['label']
     """
 
     def __init__(self, raw_dir=None, random_seed=717, train_size=0.7, val_size=0.1):

python/dgl/transform.py

@@ -122,7 +122,7 @@ def knn_graph(x, k, algorithm='bruteforce-blas', dist='euclidean'):
           This method is suitable for low-dimensional data (e.g. 3D
           point clouds)
 
-        * 'nn-descent' is a approximate approach from paper
+        * 'nn-descent' is an approximate approach from paper
           `Efficient k-nearest neighbor graph construction for generic similarity
           measures <https://www.cs.princeton.edu/cass/papers/www11.pdf>`_. This method
           will search for nearest neighbor candidates in "neighbors' neighbors".
@@ -156,7 +156,7 @@ def knn_graph(x, k, algorithm='bruteforce-blas', dist='euclidean'):
     ...                   [0.3, 0.2, 0.4]])
     >>> knn_g = dgl.knn_graph(x, 2)  # Each node has two predecessors
     >>> knn_g.edges()
-    >>> (tensor([0, 1, 2, 2, 2, 3, 3, 3]), tensor([0, 1, 1, 2, 3, 0, 2, 3]))
+    (tensor([0, 1, 2, 2, 2, 3, 3, 3]), tensor([0, 1, 1, 2, 3, 0, 2, 3]))
 
     When :attr:`x` is a 3D tensor, DGL constructs multiple KNN graphs and
     and then composes them into a graph of multiple connected components.
@@ -297,7 +297,7 @@ def segmented_knn_graph(x, k, segs, algorithm='bruteforce-blas', dist='euclidean
           This method is suitable for low-dimensional data (e.g. 3D
           point clouds)
 
-        * 'nn-descent' is a approximate approach from paper
+        * 'nn-descent' is an approximate approach from paper
           `Efficient k-nearest neighbor graph construction for generic similarity
           measures <https://www.cs.princeton.edu/cass/papers/www11.pdf>`_. This method
           will search for nearest neighbor candidates in "neighbors' neighbors".
@@ -533,7 +533,7 @@ def knn(k, x, x_segs, y=None, y_segs=None, algorithm='bruteforce', dist='euclide
           This method is suitable for low-dimensional data (e.g. 3D
           point clouds)
 
-        * 'nn-descent' is a approximate approach from paper
+        * 'nn-descent' is an approximate approach from paper
           `Efficient k-nearest neighbor graph construction for generic similarity
           measures <https://www.cs.princeton.edu/cass/papers/www11.pdf>`_. This method
           will search for nearest neighbor candidates in "neighbors' neighbors".