004 Datenverarbeitung; Informatik
Refine
Year of publication
Document Type
- Article (331)
- Monograph/Edited Volume (165)
- Doctoral Thesis (157)
- Conference Proceeding (52)
- Postprint (50)
- Master's Thesis (10)
- Other (7)
- Preprint (3)
- Part of a Book (2)
- Bachelor Thesis (1)
Language
- English (586)
- German (192)
- Multiple languages (2)
Keywords
- Informatik (21)
- machine learning (17)
- Didaktik (15)
- Hochschuldidaktik (14)
- Ausbildung (13)
- answer set programming (13)
- Cloud Computing (11)
- cloud computing (11)
- Hasso-Plattner-Institut (10)
- Hasso Plattner Institute (9)
Institute
- Institut für Informatik und Computational Science (270)
- Hasso-Plattner-Institut für Digital Engineering gGmbH (215)
- Hasso-Plattner-Institut für Digital Engineering GmbH (127)
- Extern (65)
- Fachgruppe Betriebswirtschaftslehre (27)
- Mathematisch-Naturwissenschaftliche Fakultät (24)
- Wirtschaftswissenschaften (19)
- Institut für Mathematik (16)
- Bürgerliches Recht (12)
- Digital Engineering Fakultät (8)
Decubitus is one of the most relevant diseases in nursing and the most expensive to treat. It is caused by sustained pressure on tissue, so it particularly affects bed-bound patients. This work lays a foundation for pressure mattress-based decubitus prophylaxis by implementing a solution to the single-frame 2D Human Pose Estimation problem.
For this, methods of Deep Learning are employed. Two approaches are examined, a coarse-to-fine Convolutional Neural Network for direct regression of joint coordinates and a U-Net for the derivation of probability distribution heatmaps.
We conclude that training our models on a combined dataset of the publicly available Bodies at Rest and SLP data yields the best results. Furthermore, various preprocessing techniques are investigated, and a hyperparameter optimization is performed to discover an improved model architecture.
Another finding indicates that the heatmap-based approach outperforms direct regression.
This model achieves a mean per-joint position error of 9.11 cm for the Bodies at Rest data and 7.43 cm for the SLP data.
We find that it generalizes well on data from mattresses other than those seen during training but has difficulties detecting the arms correctly.
Additionally, we give a brief overview of the medical data annotation tool annoto we developed in the bachelor project and furthermore conclude that the Scrum framework and agile practices enhanced our development workflow.
Digitale Technologien bieten erhebliche politische, wirtschaftliche und gesellschaftliche Chancen. Zugleich ist der Begriff digitale Souveränität zu einem Leitmotiv im deutschen Diskurs über digitale Technologien geworden: das heißt, die Fähigkeit des Staates, seine Verantwortung wahrzunehmen und die Befähigung der Gesellschaft – und des Einzelnen – sicherzustellen, die digitale Transformation selbstbestimmt zu gestalten. Exemplarisch für die Herausforderung in Deutschland und Europa, die Vorteile digitaler Technologien zu nutzen und gleichzeitig Souveränitätsbedenken zu berücksichtigen, steht der Bildungssektor. Er umfasst Bildung als zentrales öffentliches Gut, ein schnell aufkommendes Geschäftsfeld und wachsende Bestände an hochsensiblen personenbezogenen Daten. Davon ausgehend beschreibt der Bericht Wege zur Entschärfung des Spannungsverhältnisses zwischen Digitalisierung und Souveränität auf drei verschiedenen Ebenen – Staat, Wirtschaft und Individuum – anhand konkreter technischer Projekte im Bildungsbereich: die HPI Schul-Cloud (staatliche Souveränität), die MERLOT-Datenräume (wirtschaftliche Souveränität) und die openHPI-Plattform (individuelle Souveränität).
Digital technology offers significant political, economic, and societal opportunities. At the same time, the notion of digital sovereignty has become a leitmotif in German discourse: the state’s capacity to assume its responsibilities and safeguard society’s – and individuals’ – ability to shape the digital transformation in a self-determined way. The education sector is exemplary for the challenge faced by Germany, and indeed Europe, of harnessing the benefits of digital technology while navigating concerns around sovereignty. It encompasses education as a core public good, a rapidly growing field of business, and growing pools of highly sensitive personal data. The report describes pathways to mitigating the tension between digitalization and sovereignty at three different levels – state, economy, and individual – through the lens of concrete technical projects in the education sector: the HPI Schul-Cloud (state sovereignty), the MERLOT data spaces (economic sovereignty), and the openHPI platform (individual sovereignty).
Invention
(2023)
This entry addresses invention from five different perspectives: (i) definition of the term, (ii) mechanisms underlying invention processes, (iii) (pre-)history of human inventions, (iv) intellectual property protection vs open innovation, and (v) case studies of great inventors. Regarding the definition, an invention is the outcome of a creative process taking place within a technological milieu, which is recognized as successful in terms of its effectiveness as an original technology. In the process of invention, a technological possibility becomes realized. Inventions are distinct from either discovery or innovation. In human creative processes, seven mechanisms of invention can be observed, yielding characteristic outcomes: (1) basic inventions, (2) invention branches, (3) invention combinations, (4) invention toolkits, (5) invention exaptations, (6) invention values, and (7) game-changing inventions. The development of humanity has been strongly shaped by inventions ever since early stone tools and the conception of agriculture. An “explosion of creativity” has been associated with Homo sapiens, and inventions in all fields of human endeavor have followed suit, engendering an exponential growth of cumulative culture. This culture development emerges essentially through a reuse of previous inventions, their revision, amendment and rededication. In sociocultural terms, humans have increasingly regulated processes of invention and invention-reuse through concepts such as intellectual property, patents, open innovation and licensing methods. Finally, three case studies of great inventors are considered: Edison, Marconi, and Montessori, next to a discussion of human invention processes as collaborative endeavors.
The Security Operations Center (SOC) represents a specialized unit responsible for managing security within enterprises. To aid in its responsibilities, the SOC relies heavily on a Security Information and Event Management (SIEM) system that functions as a centralized repository for all security-related data, providing a comprehensive view of the organization's security posture. Due to the ability to offer such insights, SIEMS are considered indispensable tools facilitating SOC functions, such as monitoring, threat detection, and incident response.
Despite advancements in big data architectures and analytics, most SIEMs fall short of keeping pace. Architecturally, they function merely as log search engines, lacking the support for distributed large-scale analytics. Analytically, they rely on rule-based correlation, neglecting the adoption of more advanced data science and machine learning techniques.
This thesis first proposes a blueprint for next-generation SIEM systems that emphasize distributed processing and multi-layered storage to enable data mining at a big data scale. Next, with the architectural support, it introduces two data mining approaches for advanced threat detection as part of SOC operations.
First, a novel graph mining technique that formulates threat detection within the SIEM system as a large-scale graph mining and inference problem, built on the principles of guilt-by-association and exempt-by-reputation. The approach entails the construction of a Heterogeneous Information Network (HIN) that models shared characteristics and associations among entities extracted from SIEM-related events/logs. Thereon, a novel graph-based inference algorithm is used to infer a node's maliciousness score based on its associations with other entities in the HIN. Second, an innovative outlier detection technique that imitates a SOC analyst's reasoning process to find anomalies/outliers. The approach emphasizes explainability and simplicity, achieved by combining the output of simple context-aware univariate submodels that calculate an outlier score for each entry.
Both approaches were tested in academic and real-world settings, demonstrating high performance when compared to other algorithms as well as practicality alongside a large enterprise's SIEM system.
This thesis establishes the foundation for next-generation SIEM systems that can enhance today's SOCs and facilitate the transition from human-centric to data-driven security operations.
Virtual reality can have advantages for education and learning. However, it must be adequately designed so that the learner benefits from the technological possibilities. Understanding the underlying effects of the virtual learning environment and the learner’s prior experience with virtual reality or prior knowledge of the content is necessary to design a proper virtual learning environment. This article presents a pre-study testing the design of a virtual learning environment for engineering vocational training courses. In the pre-study, 12 employees of two companies joined the training course in one of the two degrees of immersion (desktop VR and VR HMD). Quantitative results on learning success, cognitive load, usability, and motivation and qualitative learning process data were presented. The qualitative data assessment shows that overall, the employees were satisfied with the learning environment regardless of the level of immersion and that the participants asked for more guidance and structure accompanying the learning process. Further research is needed to test for solid group differences.
Like conventional software projects, projects in model-driven software engineering require adequate management of multiple versions of development artifacts, importantly allowing living with temporary inconsistencies. In the case of model-driven software engineering, employed versioning approaches also have to handle situations where different artifacts, that is, different models, are linked via automatic model transformations.
In this report, we propose a technique for jointly handling the transformation of multiple versions of a source model into corresponding versions of a target model, which enables the use of a more compact representation that may afford improved execution time of both the transformation and further analysis operations. Our approach is based on the well-known formalism of triple graph grammars and a previously introduced encoding of model version histories called multi-version models. In addition to showing the correctness of our approach with respect to the standard semantics of triple graph grammars, we conduct an empirical evaluation that demonstrates the potential benefit regarding execution time performance.
Lehrkräfte aller Fächer benötigen informatische Kompetenzen, um der wachsenden Alltagsrelevanz von Informatik und aktuell gültigen Lehrplänen gerecht zu werden. Beispielsweise verweist in Sachsen der Lehrplan für das Fach Gemeinschaftskunde, Rechtserziehung und Wirtschaft am Gymnasium mit dem für die Jahrgangsstufe 11 vorgesehenem Thema „Digitalisierung und sozialer Wandel“ auf Künstliche Intelligenz (KI) und explizit auf die Bedeutung der informatischen Bildung. Um die nötigen informatischen Grundlagen zu vermitteln, wurde für Lehramtsstudierende des Faches Politik ein Workshop erarbeitet, der die Grundlagen der Funktionsweise von KI anhand von überwachtem maschinellen Lernen in neuronalen Netzen vermittelt. Inhalt des Workshops ist es, mit Bezug auf gesellschaftliche Implikationen wie Datenschutz bei Trainingsdaten und algorithmic bias einen informierten Diskurs zu politischen Themen zu ermöglichen. Ziele des Workshops für Lehramtsstudierende mit dem Fach Politik sind: (1) Aufbau informatischer Kompetenzen in Bezug zum Thema KI, (2) Stärkung der Diskussionsfähigkeiten der Studierenden durch passende informatische Kompetenzen und (3) Anregung der Studierenden zum Transfer auf passende Themenstellungen im Politikunterricht. Das Evaluationskonzept umfasst eine Pre-Post-Befragung zur Zuversicht zur Vermittlungskompetenz unter Bezug auf maschinelles Lernen in neuronalen Netzen im Unterricht, sowie die Analyse einer abschließenden Diskussion. Für die Pre-Post-Befragung konnte eine Steigerung der Zuversicht zur Vermittlungskompetenz beobachtet werden. Die Analyse der Diskussion zeigte das Bewusstsein der Alltagsrelevanz des Themas KI bei den Teilnehmenden, aber noch keine Anwendung der informatischen Inhalte des Workshops zur Stützung der Argumente in der Diskussion.
Viele Informatikstudierende sammeln bereits vor ihrem Studium berufliche Erfahrungen im Informatikbereich, ohne dass diese inhaltlich und didaktisch im Studium berücksichtigt werden. Dieser Beitrag geht der Frage nach, welche Kompetenzen aus beruflichen Vorqualifikationen bei Informatikstudierenden existieren und wie diese in Bezug zu Anerkennungsoptionen gesetzt werden können. Betrachtet werden: die pauschale Anerkennung, die auf erworbenen Zertifikaten beruht; die individuelle Anerkennung, bei der individuell erworbene Kompetenzen nachgewiesen werden; die Adaption von individuellen Lernwegen, die Teilkompetenzen der Studierenden berücksichtigt. Es wird eine Interviewstudie vorgestellt, in der Kompetenzen für ein Sample von Informatikstudierenden mit Vorqualifikation als Fachinformatiker/in erhoben und eine Zuordnung zu den Anerkennungsoptionen vorgenommen wurde. Für die präzisere Gestaltung von Anerkennungsprozessen und zur kritischen Reflexion der eingesetzten hochschuldidaktischen Konzepte wurde eine empirische Basis geschaffen. Die vorhandenen Konzepte richten sich traditionell an Abiturienten/ innen mit sehr geringem Informatikhintergrund und berücksichtigen die tatsächlich existierende Heterogenität der Studienanfänger/innen nicht angemessen. Die Ergebnisse zeigen, dass die Befragten aus ihrer Vorqualifikation relevante fachliche Kompetenzen mitbringen, die mit den Anerkennungsoptionen korrespondieren und deren Weiterentwicklung dienen können. Darüber hinaus werden aus überfachlichen Kompetenzen wie Selbststeuerungskompetenzen weitere Erkenntnisse zur Studiengestaltung gewonnen.
Peer-Reviews werden seit geraumer Zeit in unterschiedlichen Lehrszenarien eingesetzt. In diesem Paper wird untersucht, inwieweit das Peer- Review die Auseinandersetzung mit den Inhalten eines Grundlagenmoduls in einem präsenzfreien Lehrszenario befördern kann. Dabei scheint in den Ergebnissen die Qualität der selbst erstellten Reviews einer der wichtigsten Einflussfaktoren für den Lernerfolg zu sein, während Experten-Feedback und weitere Faktoren deutlich untergeordnet erscheinen. Die Fähigkeit ausführliche Peer-Reviews zu verfassen geht einher mit dem Erwerb von fachlicher Kompetenz bzw. entsprechenden fachlichen Vorkenntnissen.
Adaption von Lernwegen in adaptierten Lehrmaterialien für Studierende mit Berufsausbildungsabschluss
(2023)
Obwohl immer mehr Menschen nicht direkt ein Studium aufnehmen, sondern zuvor eine berufliche Ausbildung absolvieren, werden die in der Ausbildung erworbenen Kompetenzen von den Hochschulen inhaltlich und didaktisch meist ignoriert. Ein Ansatz, diese Kompetenzen zu würdigen, ist die formale Anrechnung von mitgebrachten Kompetenzen als (für den Studienabschluss erforderliche) Leistungspunkte. Eine andere Variante ist der Einsatz von speziell für die Zielgruppe der Studierenden mit Vorkenntnissen adaptiertem Lehr-Lernmaterial. Um darüber hinaus individuelle Unterschiede zu berücksichtigen, erlaubt eine weitere Adaption individueller Lernpfade den Lernenden, genau die jeweils fehlenden Kompetenzen zu erwerben. In diesem Beitrag stellen wir die exemplarische Entwicklung derartigen Materials anhand des Kurses „Datenbanken“ für die Zielgruppe der Studierenden mit einer abgeschlossenen Ausbildung zum Fachinformatiker bzw. zur Fachinformatikerin vor.
Barrierefreiheit kann durch Methoden der Informatik hergestellt und ausgebaut werden. Dieser eingeladene Beitrag stellt die Anforderungen von Menschen mit den umfangreichsten Benutzererfordernissen an Software vor, die z. B. eigene Schriftsysteme wie Braille und entsprechende taktile Ausgabegeräte verwenden. Assistive Technologien umfassen dabei auch Software verschiedenster Art. Es werden die wichtigsten Kompetenzen dafür vorgestellt. Im Curriculum der Informatik können diese Kompetenzen im Rahmen von speziellen Vorlesungen und Übungen vermittelt werden oder sie werden in die jeweiligen Fachgebiete integriert. Um den Studienbetrieb ebenfalls barrierefrei zu gestalten, sind weitere Anstrengungen notwendig, die Lehrende, Verwaltung und die Hochschulleitung einbeziehen.
Eine übliche Erzählung verknüpft lange Studienzeiten und hohe Abbrecherquoten im Informatikstudium zum einen mit der sehr gut bezahlten Nebentätigkeit von Studierenden in der Informatikbranche, die deutlich studienzeitverlängernd sei; zum anderen werde wegen des hohen Bedarfs an Informatikern ein formeller Studienabschluss von den Studierenden häufig als entbehrlich betrachtet und eine Karriere in der Informatikbranche ohne abgeschlossenes Studium begonnen. In dieser Studie, durchgeführt an der Universität Potsdam, untersuchen wir, wie viele Informatikstudierende neben dem Studium innerhalb und außerhalb der Informatikbranche arbeiten, welche Erwartungen sie neben der Bezahlung damit verbinden und wie sich die Tätigkeit auf ihr Studium und ihre spätere berufliche Perspektive auswirkt. Aus aktuellem Anlass interessieren uns auch die Auswirkungen der Covid-19-Pandemie auf die Arbeitstätigkeiten der Informatikstudierenden.
The management of knowledge in organizations considers both established long-term
processes and cooperation in agile project teams. Since knowledge can be both tacit and explicit, its transfer from the individual to the organizational knowledge base poses a challenge in organizations. This challenge increases when the fluctuation of knowledge carriers is exceptionally high. Especially in large projects in which external consultants are involved, there is a risk that critical, company-relevant knowledge generated in the project will leave the company with the external knowledge carrier and thus be lost. In this paper, we show the advantages of an early warning system for knowledge management to avoid this loss. In particular, the potential of visual analytics in the context of knowledge management systems is presented and discussed. We present a project for the development of a business-critical software system and discuss the first implementations and results.
This paper presents a methodological and conceptual replication of Stieglitz and Dang-Xuan’s (2013) investigation of the role of sentiment in information-sharing behavior on social media. Whereas Stieglitz and Dang-Xuan (2013) focused on Twitter communication prior to the state parliament elections in the German states Baden-Wurttemberg, Rheinland-Pfalz, and Berlin in 2011, we test their theoretical propositions in the context of the state parliament elections in Saxony-Anhalt (Germany) 2021. We confirm the positive link between sentiment in a political Twitter message and its number of retweets in a methodological replication. In a conceptual replication, where sentiment was assessed with the alternative dictionary-based tool LIWC, the sentiment was negatively associated with the retweet volume. In line with the original study, the strength of association between sentiment and retweet time lag insignificantly differs between tweets with negative sentiment and tweets with positive sentiment. We also found that the number of an author’s followers was an essential determinant of sharing behavior. However, two hypotheses supported in the original study did not hold for our sample. Precisely, the total amount of sentiments was insignificantly linked to the time lag to the first retweet. Finally, in our data, we do not observe that the association between the overall sentiment and retweet quantity is stronger for tweets with negative sentiment than for those with positive sentiment.
Local laws on urban policy, i.e., ordinances directly affect our daily life in various ways (health, business etc.), yet in practice, for many citizens they remain impervious and complex. This article focuses on an approach to make urban policy more accessible and comprehensible to the general public and to government officials, while also addressing pertinent social media postings. Due to the intricacies of the natural language, ranging from complex legalese in ordinances to informal lingo in tweets, it is practical to harness human judgment here. To this end, we mine ordinances and tweets via reasoning based on commonsense knowledge so as to better account for pragmatics and semantics in the text. Ours is pioneering work in ordinance mining, and thus there is no prior labeled training data available for learning. This gap is filled by commonsense knowledge, a prudent choice in situations involving a lack of adequate training data. The ordinance mining can be beneficial to the public in fathoming policies and to officials in assessing policy effectiveness based on public reactions. This work contributes to smart governance, leveraging transparency in governing processes via public involvement. We focus significantly on ordinances contributing to smart cities, hence an important goal is to assess how well an urban region heads towards a smart city as per its policies mapping with smart city characteristics, and the corresponding public satisfaction.
To manage tabular data files and leverage their content in a given downstream task, practitioners often design and execute complex transformation pipelines to prepare them. The complexity of such pipelines stems from different factors, including the nature of the preparation tasks, often exploratory or ad-hoc to specific datasets; the large repertory of tools, algorithms, and frameworks that practitioners need to master; and the volume, variety, and velocity of the files to be prepared. Metadata plays a fundamental role in reducing this complexity: characterizing a file assists end users in the design of data preprocessing pipelines, and furthermore paves the way for suggestion, automation, and optimization of data preparation tasks.
Previous research in the areas of data profiling, data integration, and data cleaning, has focused on extracting and characterizing metadata regarding the content of tabular data files, i.e., about the records and attributes of tables. Content metadata are useful for the latter stages of a preprocessing pipeline, e.g., error correction, duplicate detection, or value normalization, but they require a properly formed tabular input. Therefore, these metadata are not relevant for the early stages of a preparation pipeline, i.e., to correctly parse tables out of files. In this dissertation, we turn our focus to what we call the structure of a tabular data file, i.e., the set of characters within a file that do not represent data values but are required to parse and understand the content of the file. We provide three different approaches to represent file structure, an explicit representation based on context-free grammars; an implicit representation based on file-wise similarity; and a learned representation based on machine learning.
In our first contribution, we use the grammar-based representation to characterize a set of over 3000 real-world csv files and identify multiple structural issues that let files deviate from the csv standard, e.g., by having inconsistent delimiters or containing multiple tables. We leverage our learnings about real-world files and propose Pollock, a benchmark to test how well systems parse csv files that have a non-standard structure, without any previous preparation. We report on our experiments on using Pollock to evaluate the performance of 16 real-world data management systems.
Following, we characterize the structure of files implicitly, by defining a measure of structural similarity for file pairs. We design a novel algorithm to compute this measure, which is based on a graph representation of the files' content. We leverage this algorithm and propose Mondrian, a graphical system to assist users in identifying layout templates in a dataset, classes of files that have the same structure, and therefore can be prepared by applying the same preparation pipeline.
Finally, we introduce MaGRiTTE, a novel architecture that uses self-supervised learning to automatically learn structural representations of files in the form of vectorial embeddings at three different levels: cell level, row level, and file level. We experiment with the application of structural embeddings for several tasks, namely dialect detection, row classification, and data preparation efforts estimation.
Our experimental results show that structural metadata, either identified explicitly on parsing grammars, derived implicitly as file-wise similarity, or learned with the help of machine learning architectures, is fundamental to automate several tasks, to scale up preparation to large quantities of files, and to provide repeatable preparation pipelines.
Openness indicators for the evaluation of digital platforms between the launch and maturity phase
(2024)
In recent years, the evaluation of digital platforms has become an important focus in the field of information systems science. The identification of influential indicators that drive changes in digital platforms, specifically those related to openness, is still an unresolved issue. This paper addresses the challenge of identifying measurable indicators and characterizing the transition from launch to maturity in digital platforms. It proposes a systematic analytical approach to identify relevant openness indicators for evaluation purposes. The main contributions of this study are the following (1) the development of a comprehensive procedure for analyzing indicators, (2) the categorization of indicators as evaluation metrics within a multidimensional grid-box model, (3) the selection and evaluation of relevant indicators, (4) the identification and assessment of digital platform architectures during the launch-to-maturity transition, and (5) the evaluation of the applicability of the conceptualization and design process for digital platform evaluation.
Column-oriented database systems can efficiently process transactional and analytical queries on a single node. However, increasing or peak analytical loads can quickly saturate single-node database systems. Then, a common scale-out option is using a database cluster with a single primary node for transaction processing and read-only replicas. Using (the naive) full replication, queries are distributed among nodes independently of the accessed data. This approach is relatively expensive because all nodes must store all data and apply all data modifications caused by inserts, deletes, or updates.
In contrast to full replication, partial replication is a more cost-efficient implementation: Instead of duplicating all data to all replica nodes, partial replicas store only a subset of the data while being able to process a large workload share. Besides lower storage costs, partial replicas enable (i) better scaling because replicas must potentially synchronize only subsets of the data modifications and thus have more capacity for read-only queries and (ii) better elasticity because replicas have to load less data and can be set up faster. However, splitting the overall workload evenly among the replica nodes while optimizing the data allocation is a challenging assignment problem.
The calculation of optimized data allocations in a partially replicated database cluster can be modeled using integer linear programming (ILP). ILP is a common approach for solving assignment problems, also in the context of database systems. Because ILP is not scalable, existing approaches (also for calculating partial allocations) often fall back to simple (e.g., greedy) heuristics for larger problem instances. Simple heuristics may work well but can lose optimization potential.
In this thesis, we present optimal and ILP-based heuristic programming models for calculating data fragment allocations for partially replicated database clusters. Using ILP, we are flexible to extend our models to (i) consider data modifications and reallocations and (ii) increase the robustness of allocations to compensate for node failures and workload uncertainty. We evaluate our approaches for TPC-H, TPC-DS, and a real-world accounting workload and compare the results to state-of-the-art allocation approaches. Our evaluations show significant improvements for varied allocation’s properties: Compared to existing approaches, we can, for example, (i) almost halve the amount of allocated data, (ii) improve the throughput in case of node failures and workload uncertainty while using even less memory, (iii) halve the costs of data modifications, and (iv) reallocate less than 90% of data when adding a node to the cluster. Importantly, we can calculate the corresponding ILP-based heuristic solutions within a few seconds. Finally, we demonstrate that the ideas of our ILP-based heuristics are also applicable to the index selection problem.
Knowledge about causal structures is crucial for decision support in various domains. For example, in discrete manufacturing, identifying the root causes of failures and quality deviations that interrupt the highly automated production process requires causal structural knowledge. However, in practice, root cause analysis is usually built upon individual expert knowledge about associative relationships. But, "correlation does not imply causation", and misinterpreting associations often leads to incorrect conclusions. Recent developments in methods for causal discovery from observational data have opened the opportunity for a data-driven examination. Despite its potential for data-driven decision support, omnipresent challenges impede causal discovery in real-world scenarios. In this thesis, we make a threefold contribution to improving causal discovery in practice.
(1) The growing interest in causal discovery has led to a broad spectrum of methods with specific assumptions on the data and various implementations. Hence, application in practice requires careful consideration of existing methods, which becomes laborious when dealing with various parameters, assumptions, and implementations in different programming languages. Additionally, evaluation is challenging due to the lack of ground truth in practice and limited benchmark data that reflect real-world data characteristics.
To address these issues, we present a platform-independent modular pipeline for causal discovery and a ground truth framework for synthetic data generation that provides comprehensive evaluation opportunities, e.g., to examine the accuracy of causal discovery methods in case of inappropriate assumptions.
(2) Applying constraint-based methods for causal discovery requires selecting a conditional independence (CI) test, which is particularly challenging in mixed discrete-continuous data omnipresent in many real-world scenarios. In this context, inappropriate assumptions on the data or the commonly applied discretization of continuous variables reduce the accuracy of CI decisions, leading to incorrect causal structures.
Therefore, we contribute a non-parametric CI test leveraging k-nearest neighbors methods and prove its statistical validity and power in mixed discrete-continuous data, as well as the asymptotic consistency when used in constraint-based causal discovery. An extensive evaluation of synthetic and real-world data shows that the proposed CI test outperforms state-of-the-art approaches in the accuracy of CI testing and causal discovery, particularly in settings with low sample sizes.
(3) To show the applicability and opportunities of causal discovery in practice, we examine our contributions in real-world discrete manufacturing use cases. For example, we showcase how causal structural knowledge helps to understand unforeseen production downtimes or adds decision support in case of failures and quality deviations in automotive body shop assembly lines.