Module influence_max 

Influence Maximization algorithms

This module provides algorithms for finding the top-k seed nodes that maximise information spread under a given diffusion model:

Function	Description
greedy_influence_max	Greedy hill-climbing with Monte-Carlo estimates (Kempe 2003)
celf_influence_max	CELF – lazy evaluation cuts MC calls dramatically
celf_plus_plus	CELF++ – one additional optimisation over CELF
degree_heuristic	Fast O(n log n) heuristic: pick highest-degree nodes
pagerank_heuristic	PageRank-based seed selection (directed influence proxy)

References

• Kempe, Kleinberg & Tardos (2003) – KDD 2003
• Leskovec et al. (2007) – CELF, KDD 2007
• Goyal, Lu & Lakshmanan (2011) – CELF++, WWW 2011

Structs

InfluenceMaxConfig

Configuration for influence maximization algorithms.

InfluenceMaxResult

Result returned by influence maximization routines.

Enums

DiffusionModel

Selector for which diffusion model to use during IM.

Functions

celf_influence_max

CELF influence maximization (lazy evaluation).

celf_plus_plus

CELF++ influence maximization.

degree_heuristic

High-degree seed selection heuristic.

greedy_influence_max

Greedy influence maximization using Monte-Carlo spread estimates.

pagerank_heuristic

PageRank-based seed selection heuristic.