A learning-based approach to estimate statistics of operators in continuous queries

摘要: Statistic estimation such as output size of operators is a well-studied subject in the database research community, mainly for purpose query optimization. The assumption, however, that queries are ad-hoc and therefore emphasis has been on capturing data distribution. When long standing continuous changing concerned, more direct approach, namely building an model each operator, possible. In this paper, we propose novel learning-based method. Our method consists two steps. first step to design dedicated feature extraction algorithm can be used incrementally obtain values from underlying data. second use mining generate based extracted historical To illustrate paper studies case similarity-based searches over streaming time series. Experimental results show approach provides accurate statistic estimates with low overhead.