摘要
期货市场中的关联交易是金融监管的核心问题之一。批量交易数据条件下,关联交易行为发现的难点是如何在有限时间和空间条件下快速发现违规行为。针对该问题,提出一种将交易数据的稀疏相似性矩阵映射到十字链表的聚类算法。通过在真实交易数据下的实验分析,验证该算法能够有效解决关联交易行为的发现问题;且减少了扫描数据库的次数,降低了传统算法的时间复杂度和空间复杂度。
Futures market related transaction is one of the main problems in the financial regulation. The difficulties of finding related transaction in the condition of massive transactions is how to discover irregularities quickly in the limited time and space conditions. To solve this problem, a clustering algorithm is proposed, which is based on sparse similarity matrix of transaction data mapped to cross list. It is verified by experiments in a real transaction data that the algorithm can effectively solve the problems found in the related transactions, reducing time complexity and space complexity of traditional algorithm.
出处
《计算机应用与软件》
CSCD
2016年第12期275-278,294,共5页
Computer Applications and Software
关键词
期货市场
关联交易
十字链表
聚类分析
Futures market
Related transaction
Orthogonal list
Clustering
