Search systems are a crucial means for users to access information. However, considering only the top search results naturally comes with biases that increase even further with personalization, biased databases, or intransparent retrieval systems. We believe that it is essential that users can a) easily understand the characteristics of their search results and b) control them. Our work investigates how the demand for a more fair attribute representation or more diverse perspectives on a topic can be reflected in search results to various degrees. We propose re-ranking methods integrating those user-defined variety settings and an extension of a diversity evaluation metric regarding exposure. Applied to literature search in the multidisciplinary domain of water the methods effectively adjust rankings to match different variety preferences. With that, we demonstrate the potential to enhance user control over search variety, contributing to transparency about entrenched biases and promoting a more user-centered search experience.

Question answering (QA) frameworks typically assume that a user input solely consists of a question, including perhaps a preceding sentence. In several domains, however, this assumption is not realistic and may lead to poor quality of generated or retrieved answers. This holds in particular true for the legal and medical domains, where users sometimes provide a substantial amount of context information before posing a question, all in a single input.

In this paper, we investigate a QA framework in the medical domain that is agnostic to facts-of-a-case (FoC), often very elaborate statements made by users to describe a health-related context before posing a question. For this, we present a novel German QA-dataset from the medical domain that has such real life medical questions as well as answers from experts. Based on that data, we investigate different approaches to process the FoC from a question and process them using state-of-the-art open source and commercial large language models. The results reveal interesting insights into proper approaches to effectively deal with FoC type of questions, ensuring high quality answers.

Data profiling is the process of extracting implicit metadata, such as data types, attribute statistics, and various types of data dependencies, from raw datasets. Because structural metadata is often not stored explicitly, many data management and data engineering applications rely on data profiling to identify, for example, functional dependencies, inclusion dependencies, or uniqueness column combinations. The utilization of automatically discovered metadata within use cases, such as data discovery, integration, normalization, cleaning, or optimization, is, however, a still very complicated process. This is because existing data profiling approaches consider different types of metadata in isolation, although in practice, many use cases require specific combinations, i.e., patterns of structural metadata. For this reason, we recently proposed the Data Profiling Query Language DPQL that can express (and in the near future holistically discover) arbitrary patterns of various types of metadata. DPQL query patterns allow data scientists to express exactly what metadata real-world applications require. This eliminates otherwise complicated post-processing efforts, reduces result sizes, and might lead to significantly shorter profiling times. To demonstrate the expressiveness and versatility of DPQL, we survey a variety of data engineering applications in this paper and solve their metadata requirements with concrete DPQL query patterns. We also measure DPQL result sizes on two benchmark datasets and compare them to the result sizes of standard data profiling algorithms to demonstrate that holistic data profiling with DPQL produces not only more suitable, but also much smaller result sets.