node2vec

library0.5.0pypypi✓ verified 83d ago

Python implementation of the node2vec algorithm for learning continuous feature representations for nodes in networks. Version 0.5.0 supports Python 3.8-3.11, with sparse matrix input and optional GPU acceleration via CuPy.

ai-ml

pip install node2vec

github ↗homepage ↗

INSTALL

IMPORT

SIG · NODE2VEC

node2vec

ai-mlpythonv0.5.0

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

Node2Vec

✓ from node2vec import Node2Vec

✗ from node2vec.node2vec import Node2Vec

Internal module path changed in 0.4.0; top-level import is recommended

Node2Vec

✓ from node2vec import Node2Vec

✗ import node2vec; model = node2vec.Node2Vec(...)

Import the class directly to avoid attribute errors

Basic usage: create a graph, generate walks, and train embeddings.

import networkx as nx from node2vec import Node2Vec # Create a graph g = nx.karate_club_graph() # Precompute probabilities and generate walks node2vec = Node2Vec(g, dimensions=64, walk_length=30, num_walks=200, workers=4) # Train Word2Vec model model = node2vec.fit(window=10, min_count=1, batch_words=4) # Get embedding for node 0 embedding = model.wv['0'] print(embedding.shape)

Debug

Known issues

breakingIn v0.5.0, the 'target' parameter for biased walks is removed. Use 'p' and 'q' parameters to control return and in-out parameters.

fix

Replace 'target' with 'p' and 'q' in Node2Vec constructor.

affects: >=0.5.0

gotchaNode identifiers must be strings. Passing non-string node IDs (e.g., integers) will cause a cryptic error in Gensim's Word2Vec.

fix

Convert node IDs to strings: nx.relabel_nodes(g, {n: str(n) for n in g.nodes()})

affects: all

gotchaGraph must be connected. Disconnected nodes will not appear in walks and embeddings will be missing.

fix

Ensure the graph is a single connected component or handle missing nodes after training.

affects: all

deprecatedThe 'weight_key' parameter for edge weights is deprecated and will be removed in a future version. Use 'edge_weight_attr' instead.

fix

Replace weight_key='weight' with edge_weight_attr='weight' in Node2Vec constructor.

affects: >=0.5.0

Errors

Common errors & fixes

AttributeError: module 'node2vec' has no attribute 'Node2Vec'

Importing the module instead of the class, or using outdated import path.

fix

Use 'from node2vec import Node2Vec'.

ValueError: Node '0' not in graph

Trying to get embedding for a node that doesn't exist in the graph, or node IDs were converted to strings but lookup uses original type.

fix

Ensure all node lookups use the same type (string) as the graph nodes: model.wv[str(0)].

KeyError: "Key '0' not in vocabulary"

The node did not appear in any generated walks, likely because the graph is disconnected or node degree is zero.

fix

Verify graph connectivity and ensure all nodes have at least one walk.

ImportError: cannot import name 'Node2Vec' from 'node2vec'

Installed version is older than 0.4.0, where the class was not exposed at top level.

fix

Upgrade: pip install --upgrade node2vec, then use from node2vec import Node2Vec.

Upgrade

Version history

0.5.0latest on PyPI · released Aug 2, 2024

Audit

Dependencies

numpyrequiredNumerical arrays

networkxrequiredGraph representation

gensimrequiredWord2Vec implementation used for embedding

cupy-cuda11xoptionalGPU acceleration (optional)

tqdmrequiredProgress bars

scipyrequiredSparse matrix support

Agent activity

13 hits · last 30 days

node

Resources

githubhttps://github.com/eliorc/node2vec ↗

homepagehttps://github.com/eliorc/node2vec ↗