Parallel Particle Swarm Optimization Based on Spark for Academic Paper Co-Authorship Prediction
Parallel Particle Swarm Optimization Based on Spark for Academic Paper Co-Authorship Prediction
复制标题
DOI：
10.3390/info12120530
复制
发表时间：
2021-12
期刊：
影响因子：
--
通讯作者：
Congmin Yang;Tao Zhu;Yang Zhang;Huansheng Ning;L. Chen;Zhenyu Liu
中科院分区：
文献类型：
--
作者：
Congmin Yang;Tao Zhu;Yang Zhang;Huansheng Ning;L. Chen;Zhenyu Liu
粒子群优化算法在各种优化问题中得到了广泛的应用。尽管粒子群算法在许多领域都取得了成功，但在大数据应用中解决优化问题往往需要处理大量数据，而传统粒子群算法在单机上无法处理这些数据。目前已经有一些基于Spark的并行粒子群，但它们大多是用于解决数值优化问题，而很少用于解决大数据优化问题。在本文中，我们提出了一种新的基于spark的并行粒子群算法来预测学术论文的共同作者，我们将其描述为一个来自大量学术数据的优化问题。实验结果表明，所提出的并行粒子群算法能够达到较好的预测精度。
The particle swarm optimization (PSO) algorithm has been widely used in various optimization problems. Although PSO has been successful in many fields, solving optimization problems in big data applications often requires processing of massive amounts of data, which cannot be handled by traditional PSO on a single machine. There have been several parallel PSO based on Spark, however they are almost proposed for solving numerical optimization problems, and few for big data optimization problems. In this paper, we propose a new Spark-based parallel PSO algorithm to predict the co-authorship of academic papers, which we formulate as an optimization problem from massive academic data. Experimental results show that the proposed parallel PSO can achieve good prediction accuracy.