一种基于索引的概念格分布式构造方法

Distributed algorithm for constructing concept lattice based on index

现有的概念格并行/分布式构造算法在处理较大规模数据时,需要搜索大量不相关概念,降低了算法性能。为此,提出了一种基于索引的概念格分布式构造方法——LCBI,插入新概念时先利用索引快速找出新概念的极大相关概念,再对所有极大相关概念的子概念进行自顶向下地并行搜索以找出它们的交叉子概念,从而减少了搜索范围。理论分析和实验表明,在处理大规模稠密数据时,LCBI比其他分布式算法具有较明显的优势。

The presented concept lattice parallel/distributed algorithm needs to search plenty of non-related concepts when dealing with a large scale data, which reduces the performance of the algorithm. A distributed concept lattice construction algorithm based on index named LCBI was put forward. When inserting a new concept, it quickly found all the greatest correlative concepts of the new concept using index, then found out cross-sub-concepts of child nodes of all greatest correlative concepts using parallel and top-down search, which decreased the search area. Theoretical analysis and experimental results show that LCBI outperforms the other distributed algorithms when dealing with dense context.