In this paper, we introduce STEER to adapt learned semantic type extraction approaches to a new, unseen data lake. STEER provides a data programming framework for semantic labeling which is used to generate new labeled training data with minimal overhead. At its core, STEER comes with a novel training data generation procedure called Steered-Labeling that can generate high quality training data not only for non-numeric but also for numerical columns. With this generated training data STEER is able to fine-tune existing learned semantic type extraction models. We evaluate our approach on four different data lakes and show that we can significantly improve the performance of two different types of learned models across all data lakes.

Today, developing the conceptual design of information systems has become inconceivable without the support of graphical models. Models are developed using graph-based modeling languages such as UML. This study focuses on the impact of metamodel properties on different usability attributes in the domain of graphical modeling languages. The study is based on a model of hypotheses including a structural causal model developed under consideration of cognitive theories and usability theory. Survey data is collected and the causal hypotheses are assessed using a structure equation modeling approach. The study shows important findings for practical and theoretical issues of how differing modeling languages are influencing usability attributes on causal stages in the modeling domain. The main focus of this article is to define differences and similarities of how language metaproperties impact usability attributes in model interpretation scenarios.