基于图的邻接点优先的联合树算法的研究与实现

贝叶斯网络是以概率理论为基础的不确定知识表示模型,联合树算法是一种应用广泛的贝叶斯网络推理算法。提出了基于邻接点优先的联合树算法,从图模型和计算效率两个方面对联合树算法（JT）和基于图的邻接点优先的联合树（AD-JT）算法进行推理时间的比较,实验表明：基于图的邻接点优先的联合树算法能够有效地处理大规模数据,极大地减少了消耗时间,计算效率有显著改进。

Study and Implementation of Junction Tree Algorithm based on First Search of Adjacent Node of Graph

Bayesian Network is an uncertain knowledge-representation model based on probability,while the junction tree algorithm is a widely-applied Bayesian network inference algorithm.This paper proposes a new junction tree algorithm based on first search of adjacent node of graph.Junction tree algorithm and junction tree algorithm based on first search of adjacency node of graph are compared in their inference time from the perspective of graphical structure and computational efficiency.Experimental results show that junction tree algorithm based on first search of adjacent node of graph is effective and efficient in large scale databases,could greatly reduce the time-consumption and improve the computing efficiency.