Upload README.md

README.md

@@ -1,77 +1,3 @@
-# TEG-DB
-
-This is a dataset benchmark, which contains many datasets with rich textual edges. 
-
-## File Structure
-
-- dataset_name/
-    - raw/
-       dataset.pkl: The processed graph data in '.pkl' format.
-  
-    - processed/
-      
-        download_data.sh: Script to download the raw data.
-  
-        data_processing.py: Script to process original data into 'pkl' format
-        
-        instructions.txt: Instructions on how to use the scripts.
-        
-    - embeddings/
-        text_embeddings.pt: Text embeddings generated from the raw data.
-## Usage loading raw data
-```python
-    import pickle
-    with open('data/raw/dataset.pkl', 'rb') as f:
-        data = pickle.load(f)
-```
-## Generate graph data
-Users can also generate graph data according to their own needs, either by utilizing the data processing methods provided or by creating their own graph data based on them.
-
-### Data download
-```bash
-bash dataset_name/processed/download_data.sh
-```
-### Data processing
-```bash
-python dataset_name/processed/dataset_processing.py
-```
-## Graph structure and attribute
-
-- User-Book Reviews Networks（Goodreads-):
-    
-      text_nodes = nodes class.Users can access this attribute to obtain the type of each node in the homogeneous graph. There are three types of nodes: user, book, and genres (with a total of ten types
-      nodes_text = The summary of each books
-      edge_index = Contains the node index pairs for all edges.
-      text_edges  = The reviews between users and books, which means edges textual information
-- Citation Networks 
-
-      text_nodes = The titles of papers.
-      text_node_labels = The category of each paper.
-      edge_index = Contains the node index pairs for all edges
-      text_edges = Citation information. such as citation contexts, citation paragraph, etc. 
-
-- Shopping Networks(Amazon-)
-
-      text_nodes = nodes class.Users can access this attribute to obtain the type of each node in the homogeneous graph. There are two types of nodes: users and asins(product)
-      edge_index = Contains the node index pairs for all edges.
-      text_edges = The reviews between users and products
-      node_texts = The summary of the product.
-      text_node_labels = The category of the product
-- Social Networks(Reddit and Twitter)
-
-    Twitter:
-
-       text_nodes = The text for user nodes is "user", and the text for tweet nodes is the actual tweet content.
-       text_edges = The text for user-user edges is the tweet content, and the text for user-tweet edges is an empty string.
-       node_labels = Contains the labels for all nodes. User nodes are labeled as -1, and tweet nodes are labeled with their event IDs (e.g., 1 and 2).
-       edge_labels = Contains the labels for all edges. User-user edges are labeled with the event ID of the tweet (e.g., 1 and 2), and user-tweet edges are labeled as -1.
-       edge_index = Contains the node index pairs for all edges.
-    Reddit
-        
-       text_nodes = The text for subreddit nodes is the name of the subreddit, and the text for user nodes is the user flair.
-       text_edges = The text for subreddit-user edges is the reddit comments.
-       node_labels = Contains the labels for all nodes. Subreddit nodes are labeled as -1, and user nodes are labeled by whether they are a moderator for any Reddit community (0 or 1).
-       edge_score_labels = Contains the scores for all edges. The score for a comment is upvotes minus downvotes.
-       edge_special_labels = Contains the binary labels for all edges. A common comment is labeled as 0, and a distinguished comment (sent by a moderator of a community) is labeled as 1.
-       edge_index = Contains the node index pairs for all edges.
-
+---
+license: mit
+---