New publication in the VLDB Journal

We are happy to announce that our article Efficient detection of multivariate correlations with different correlation measures has been accepted for publication in the VLDB Journal.

This article is an extended version of our VLDB 2022 paper, and includes the following:

1. An extension of the static algorithm to support two new correlation measures: Total Correlation and Euclidean Similarity.
2. A redesign of CDStream’s streaming model, enabling faster detection of concept drift and improved recall.
3. Introduction of a new routing principle for traversing the search space, causing a 2x speedup for top-k queries.
4. A new experimental evaluation of the algorithms, with more datasets, queries, and a comparison with state-of-the-art DBMS.

The article is available here.