【摘要】 针对影响力最大化问题中贪心算法时间效率低的局限性,提出基于拓扑势的影响力最大化算法。基于拓扑势理论,确定节点是"山峰""山谷"和"斜坡"节点;启发式地选取加权度最大的k个"山峰"和"山谷"节点构成候选种子集;采用CELF (cost-effective forward)算法确定最优种子集,提高影响范围。实验结果表明,基于拓扑势的算法在Amazon数据集上比贪心算法的运行时间快了98%,在时间复杂度方面比其它传统算法更具优势。更多还原

【Abstract】 To deal with the low efficiency of greedy algorithm,an algorithm for influence maximization based on topology potential was proposed.The topological potential theory was utilized to ensure the peak,valley and slope nodes.Top k weighted degree nodes were selected as candidate seed set by combining peak and valley nodes.The CELF(cost-effective forward)algorithm was conducted to ensure the optimal seed set,which improved both efficiency and accuracy.The experimental results show that this algorithm on Amazon dataset runs 98% faster than greedy algorithm,which performs better than other traditional algorithms in efficiency.更多还原