论文部分内容阅读
【目的】通过对标签传播方式的控制,提高社区发现的质量和效率,提升社区发现在推荐系统中的能力。【方法】提出一种高效的基于临近节点影响力强度的标签传播社区发现算法,利用临近节点间的影响强度优化标签的传播路径。【结果】在真实数据集和人工数据集上的实验结果表明,利用邻近节点间的相互影响强度进行标签的传播和更新,本文的算法社区发现准确率比经典LPA算法提高2-5倍,比MLPA算法提高约10%。【局限】实验数据的规模有待加强,临近节点影响强度的概念模型的推广还需要完善。【结论】为提高社区发现的质量,减少标签传播的不稳定性提供一种可行方案。
[Purpose] To improve the quality and efficiency of community discovery through the control over the way of label transmission, and to improve the ability of community discovery in recommendation system. 【Method】 An efficient label-based community discovery algorithm based on the influence intensity of neighboring nodes is proposed, and the propagation path of labels is optimized by the influence intensity of adjacent nodes. 【Result】 The experimental results on real datasets and artificial datasets show that the accuracy of the algorithm community discovery is improved by 2-5 times than that of the classical LPA algorithm by using the mutual influence intensity of adjacent nodes to transmit and update the labels. The MLPA algorithm improves by about 10%. [Limitations] The scale of the experimental data needs to be strengthened, and the promotion of the conceptual model near the node impact strength needs further improvement. 【Conclusion】 To provide a feasible solution to improve the quality of community discovery and reduce the instability of label transmission.