P311 - BTW2021- Datenbanksysteme für Business, Technologie und Web

Aggregate-based Training Phase for ML-based Cardinality Estimation

2021-03-16T07:57:13Z

Aggregate-based Training Phase for ML-based Cardinality Estimation Woltmann, Lucas; Hartmann, Claudio; Habich, Dirk; Lehner, Wolfgang Kai-Uwe Sattler; Melanie Herschel; Wolfgang Lehner Cardinality estimation is a fundamental task in database query processing and optimization. As shown in recent papers, machine learning (ML)-based approaches may deliver more accurate cardinality estimations than traditional approaches. However, a lot of training queries have to be executed during the model training phase to learn a data-dependent ML model making it very time-consuming. Many of those training or example queries use the same base data, have the same query structure, and only differ in their selective predicates. To speed up the model training phase, our core idea is to determine a predicate-independent pre-aggregation of the base data and to execute the example queries over this pre-aggregated data. Based on this idea, we present a specific aggregate-based training phase for ML-based cardinality estimation approaches in this paper. As we are going to show with different workloads in our evaluation, we are able to achieve an average speedup of 63 with our aggregate-based training phase and thus outperform indexes.

Cluster Flow - an Advanced Concept for Ensemble-Enabling, Interactive Clustering

2021-03-16T07:57:13Z

Cluster Flow - an Advanced Concept for Ensemble-Enabling, Interactive Clustering Obermeier, Sandra; Beer, Anna; Wahl, Florian; Seidl, Thomas Kai-Uwe Sattler; Melanie Herschel; Wolfgang Lehner Even though most clustering algorithms serve knowledge discovery in fields other than computer science, most of them still require users to be familiar with programming or data mining to some extent. As that often prevents efficient research, we developed an easy to use, highly explainable clustering method accompanied by an interactive tool for clustering. It is based on intuitively understandable kNN graphs and the subsequent application of adaptable filters, which can be combined ensemble-like and iteratively and prune unnecessary or misleading edges. For a first overview of the data, fully automatic predefined filter cascades deliver robust results. A selection of simple filters and combination methods that can be chosen interactively yield very good results on benchmark datasets compared to various algorithms.

Using FALCES against bias in automated decisions by integrating fairness in dynamic model ensembles

2021-03-16T07:57:13Z

Using FALCES against bias in automated decisions by integrating fairness in dynamic model ensembles Lässig, Nico; Oppold, Sarah; Herschel, Melanie Kai-Uwe Sattler; Melanie Herschel; Wolfgang Lehner As regularly reported in the media, automated classifications and decisions based on machine learning models can cause unfair treatment of certain groups of a general population. Classically, the machine learning models are designed to make highly accurate decisions in general. When one machine learning model is not sufficient to define the possibly complex boundary between classes, multiple specialized" models are used within a model ensemble to further boost accuracy. In particular

Optimized Theta-Join Processing

2021-03-16T07:57:12Z

Optimized Theta-Join Processing Weise, Julian; Schmidl, Sebastian; Papenbrock, Thorsten Kai-Uwe Sattler; Melanie Herschel; Wolfgang Lehner The Theta-Join is a powerful operation to connect tuples of different relational tables based on arbitrary conditions. The operation is a fundamental requirement for many data-driven use cases, such as data cleaning, consistency checking, and hypothesis testing. However, processing theta-joins without equality predicates is an expensive operation, because basically all database management systems (DBMSs) translate theta-joins into a Cartesian product with a post-filter for non-matching tuple pairs. This seems to be necessary, because most join optimization techniques, such as indexing, hashing, bloom-filters, or sorting, do not work for theta-joins with combinations of inequality predicates based on <, ?, ?, ?, >. In this paper, we therefore study and evaluate optimization approaches for the efficient execution of theta-joins. More specifically, we propose a theta-join algorithm that exploits the high selectivity of theta-joins to prune most join candidates early; the algorithm also parallelizes and distributes the processing (over CPU cores and compute nodes, respectively) for scalable query processing. The algorithm is baked into our distributed in-memory database system prototype A2DB. Our evaluation on various real-world and synthetic datasets shows that A2DB significantly outperforms existing single-machine DBMSs including PostgreSQL and distributed data processing systems, such as Apache SparkSQL, in processing highly selective theta-join queries.