面向高速乱序流的top-k连续查询方法

Continuous Top-k Query Method over High-speed Out-of-order Data Streams

提出一种面向高速乱序流的top-k连续查询方法。使用基于缓存的方法等待迟到元组,但不对缓冲区内数据进行排序,通过统计运行信息实现缓存时长自适应,然后使用改造的MinTopk算法计算当前窗口的top-k结果集。实验结果表明,该方法在高速乱序流上实现了高效的top-k查询,在保证用户允许的最小正确率的情况下计算出最小缓存时长,减少了查询时延。

The continuous top-k query approach over high-speed out-of-order data streams was proposed.Using a cache-based method to wait for late tuples without sorting the data in the buffer,the self-adaptive cache duration was realized by counting the running information.And the modified MinTopk algorithm was used to calculate the top-k result set of the current window.The experimental results showed that this approach could achieve efficient top-k query over high-speed out-of-order data streams.In case of ensuring the minimum accuracy allowed by users,the minimum cache duration was calculated to reduce the query delay.