Worst Case Optimal Joins on Relational and XML data

Abstract

In recent data management ecosystem, one of the greatest challenges is the data variety. Data varies in multiple formats such as relational and (semi-)structured data. Traditional database handles a single type of data format and thus its ability to deal with different types of data formats is limited. To overcome such limitation, we propose a multi-model processing framework for relational and semi-structured data (i.e. XML), and design a worst-case optimal join algorithm. The salient feature of our algorithm is that it can guarantee that the intermediate results are no larger than the worst-case join results. Preliminary results show that our multi-model algorithm significantly outperforms the baseline join methods in terms of running time and intermediate result size.