thunlp/OpenNE

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1706 stars · 482 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

OpenNE is an open-source Python toolkit for network embedding (graph representation learning) that unifies training and testing interfaces for classic NE models like DeepWalk, LINE, node2vec, GraRep, GCN, and TADW. It implements models on TensorFlow to support GPU training and includes evaluation on node classification tasks.

Use cases

• learn node embeddings for a graph dataset
• train deepwalk or node2vec on an edgelist
• compare network embedding models on node classification
• incorporate node text attributes into graph embeddings with TADW
• reproduce results from network embedding papers
• generate graph features for downstream machine learning tasks

When to choose

• you need a unified interface to many classic network embedding algorithms
• you want to benchmark NE models with standard node classification evaluation
• you need text-attributed network embedding via TADW
• you want GPU-accelerated training of NE models via TensorFlow

When to avoid

• you need modern graph neural network frameworks with dynamic or heterogeneous graph support
• you prefer PyTorch-based tooling like PyTorch Geometric or DGL
• your project requires actively maintained software with recent updates
• you need large-scale distributed graph embedding beyond the scope of this toolkit

Facets

library · maturity maintenance

machine-learning data-science cli machine-learning developer-tools python network-embedding graph-embeddings node2vec deepwalk graph-representation-learning tensorflow algorithms linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="thunlp/OpenNE")