eliorc/node2vec

Leverages biased random walks parameterized by return probability (p) and in-out parameters (q) to generate node sequences, then applies gensim's Word2Vec for embedding. Supports weighted graphs, node-specific sampling strategies, and memory-efficient processing via temporary folder storage for large networks. Provides multiple edge embedding methods (Hadamard, average, weighted L1/L2) to derive edge representations from node embeddings.

1,293 stars and 43,847 monthly downloads. No commits in the last 6 months. Available on PyPI.