abstract.tex

\begin{abstract}
    With the continuing growth of stored data, data compression has become a common task in database systems regarding query processing or optimization. Among the large variety of existing lightweight integer compression algorithms, there is no single-best one. Thus, a selection strategy for finding suitable algorithms is necessary. In addition to the compression algorithm itself, the algorithm parameterization also influences the compression results. Hence, we present a \emph{Learned Selection Strategy for Lightweight Integer Compression Algorithm Parameterizations} which extends existing Machine Learning approaches by considering both, the selection of the best-fitting algorithm and the parameterization leading to the best compression result. We evaluate our strategy against a baseline, point out advantages of our approach and explain the behavior of our Machine Learning models by analyzing feature importances. We show that the usage of our \emph{Learned Selection Strategy for Lightweight Integer Compression Algorithm Parameterizations} lead to better compression results than using the simplest algorithm with a standard parameterization. 
\end{abstract}