/BDC_GCN_LinkPrediction

Datathon2022

Primary LanguageJupyter Notebook

BDC_GCN_LinkPrediction

Datathon2022

Result

Classification accuracy (both on train and val, True Positive and True negative rate trade-off shown in result.png.

  • Data preparation: node_preprocessing.py

  • Training process runable: vis.ipynb or dgl_test.py

  • EDA: eda.ipynb

  • trained weights: model_final.pth

Data

  • nodes (number id): webpage
    • (22470 linked, 1655 isolated)
  • edge: exists if two pages link to each other (132039)
  • Page’s text description
  • Page type (label)

Pre-processing

  • Node features

    • labels: provided, 4 types
    • Embedding text one-hot vectors
    • Use Doc2Vec, decide the output feature dimension based on the raw sentence length

  • Problem Abstraction: Link Prediction in Graph

  • Future improvement: small model — room to increase complexity

    • Deeper GraphSAGE
    • higher number of channels
    • longer text embedding

Graph

  • Nodes: pages
  • Edges: connectivity of pages
  • Node feature: label + (embedded) text