一种有效的量化交易数据相似性搜索方法

An Efficient Method for Similarity Search on Quantitative Transaction Data

量化交易数据与一般交易数据的不同之处在于它在各个维上的值是数值型而不是二值型的 研究这种数据的有效的相似性搜索方法是一个重要而具有挑战性的课题 提出了一个新的相似性度量函数Hsim() ,这个度量函数可以较好地克服Lp 等传统的距离函数在高维空间中的缺点 ,并能将二值型和数值型数据距离的计算整合到一个统一的框架中去 结合量化交易数据的特点 ,构造了定义在该函数上的相似性索引结构 ,并对建立在该索引结构上的相似性查询方法进行了阐述 实验表明 ,这种搜索方法对量化交易数据的相似性搜索有较高的修剪率 。

The difference of the quantitative transaction data from the common transaction data is that the value of each dimension is quantitative, not binary. The study of the efficient method for similarity search on the quantitative transaction data is very important and challenging. A new function Hsim() is presented to measure the proximity of objects in high dimensional spaces. The function can overcome the shortcoming of L p-norm and other distance functions, and adapt to binary and numerical data. According to the characteristic of the quantitative transaction data, a similarity indexing structure based on Hsim() is constructed, and an algorithm for similarity search on quantitative transaction data is also described. Experiments demonstrate that this method has very good pruning efficiency for similarity search on the quantitative transaction data, so it can greatly speed the similarity search.