University of Konstanz
KOPS - The Institutional Repository of the University of KonstanzNot a member yet
31050 research outputs found
Sort by
From speech signal to syntactic structure : A computational implementation
No full textThis paper presents a new computational implementation bridging several modules of grammar from phonetics to phonology to syntax. The system takes as input a speech signal annotated with syllables, interprets the phonetic data in phonological/prosodic terms, matches the data against a lexicon and makes the results available to a linguistically deep computational grammar. The system is showcased by means of syntactically ambiguous structures in German which can be disambiguated based on prosodic constituency information. A system evaluation with the German data showed good results for this new combination of automatic speech signal analysis and computational grammars, which takes a significant step towards a linguistically fine-grained computational analysis and hence towards real automatic speech understanding.publishe
Improving Computer Vision Interpretability : Transparent Two-Level Classification for Complex Scenes
No full textTreating images as data has become increasingly popular in political science. While existing classifiers for images reach high levels of accuracy, it is difficult to systematically assess the visual features on which they base their classification. This paper presents a two-level classification method that addresses this transparency problem. At the first stage, an image segmenter detects the objects present in the image and a feature vector is created from those objects. In the second stage, this feature vector is used as input for standard machine learning classifiers to discriminate between images. We apply this method to a new dataset of more than 140,000 images to detect which ones display political protest. This analysis demonstrates three advantages to this paper’s approach. First, identifying objects in images improves transparency by providing human-understandable labels for the objects shown on an image. Second, knowing these objects enables analysis of which distinguish protest images from non-protest ones. Third, comparing the importance of objects across countries reveals how protest behavior varies. These insights are not available using conventional computer vision classifiers and provide new opportunities for comparative research.publishe
ENQUIRE automatically reconstructs, expands, and drives enrichment analysis of gene and Mesh co-occurrence networks from context-specific biomedical literature
No full textThe accelerating growth of scientific literature overwhelms our capacity to manually distil complex phenomena like molecular networks linked to diseases. Moreover, biases in biomedical research and database annotation limit our interpretation of facts and generation of hypotheses. ENQUIRE (Expanding Networks by Querying Unexpectedly Inter-Related Entities) offers a time- and resource-efficient alternative to manual literature curation and database mining. ENQUIRE reconstructs and expands co-occurrence networks of genes and biomedical ontologies from user-selected input corpora and network-inferred PubMed queries. Its modest resource usage and the integration of text mining, automatic querying, and network-based statistics mitigating literature biases makes ENQUIRE unique in its broad-scope applications. For example, ENQUIRE can generate co-occurrence gene networks that reflect high-confidence, functional networks. When tested on case studies spanning cancer, cell differentiation, and immunity, ENQUIRE identified interlinked genes and enriched pathways unique to each topic, thereby preserving their underlying context specificity. ENQUIRE supports biomedical researchers by easing literature annotation, boosting hypothesis formulation, and facilitating the identification of molecular targets for subsequent experimentation.publishe
Armed violent conflict and healthcare-seeking behavior for maternal and child health in sub-Saharan Africa : A systematic review
No full textBackground: Over 630 million women and children worldwide have been displaced by conflict or live dangerously close to conflict zones. While the adverse effects of physical destruction on healthcare delivery are relatively well understood, the effects on healthcare-seeking behavior remain underexplored, particularly in sub-Saharan Africa. This study aims to better understand the interconnections and knowledge gaps between exposure to armed violent conflicts and healthcare-seeking behaviors for maternal and child health in sub-Saharan Africa.
Methods: Five key electronic databases (PubMed, Scopus, Web of Science, PsycNET, and African Journals Online) were searched for peer-reviewed publications between 2000 and 2022. The review was designed according to PRISMA-P statement and the protocol was registered with PROSPERO database. The methodological quality and risks of bias were appraised using GRADE. A data extraction instrument was modelled along the Cochrane Handbook for Systematic Reviews and the Centre for Reviews and Dissemination of Systematic Reviews.
Result: The search results yielded 1,148 publications. Only twenty-one studies met the eligibility criteria, reporting healthcare-seeking behaviors for maternal and child health. Of the twenty-one studies, seventeen (81.0%) reported maternal health behaviors such as antenatal care, skilled birth attendance, postnatal care services, and family planning. Nine studies (42.8%) observed behaviors for child health such as vaccination uptake, case management for pneumonia, diarrhea, malnutrition, and cough. While conflict exposure is generally associated with less favorable healthcare-seeking behaviors, some of the studies found improved health outcomes. Marital status, male partner attitudes, education, income and poverty levels were associated with healthcare-seeking behavior.
Conclusion: There is a need for multifaceted interventions to mitigate the impact of armed violent conflict on healthcare-seeking behavior, given its overall negative effects on child and maternal healthcare utilization. While armed violent conflict disproportionately affects children’s health compared to maternal health, it is noteworthy that exposure to such conflicts may inadvertently also lead to positive outcomes.
Prospero registration number: CRD42023484004.publishe
The when and how of planning : Meta-analysis of the scope and components of implementation intentions in 642 tests
No full textWhen and how should one plan? We estimated the scope (when) of implementation intentions by computing effect sizes for different outcomes, samples, and study characteristics, and tested the components (how) of implementation intentions by analysing the format, processes of formation, and contents of plans. Across 642 independent tests, forming implementation intentions proved effective for cognitive, affective, and behavioural outcomes (.27 ≤ d ≤ .66). Effect sizes were larger when plans had a contingent (if-then) format, participants were highly motivated to pursue the goal, and plans were rehearsed. We developed a new taxonomy of the cues (e.g., time-and-place, task juncture) and responses (e.g., cognitive procedures, ignore- or inner speech-responses) specified in implementation intentions and tested their efficacy in promoting outcomes. Our review underlines the utility of implementation intentions in helping people regulate their thoughts, feelings, and actions and offers a taxonomy of plan contents that could inspire further tests of implementation intentions.publishe
Reflections on the Uses and Available Choices of Categorical Colorschemes
No full textCategorical colorschemes must respect a number of criteria — mainly, they need to incorporate a number of easily distinguishable colors, and they need to avoid giving to the reader the impression that the colors in the visualization have particular relationships. Crafting these palettes requires careful attention to the distribution of colors; thus, for a long time, visualization designers have been relying on a limited choice of readily available palettes. Although such palettes have been proven practical and functional, our own experience with designing visualizations had us struggle repeatedly with the limited choice, the feeling of repetitiveness in seeing the same colors in visualization papers, and a number of other limitations that we discuss in the paper. In this document, we discuss some properties of the most common categorical colorschemes, and propose a method to generate new palettes that are comparable in properties to the existing ones.publishe
Regaining Control : Enabling Educators to Build Specialized AI Chat Bots with Retrieval Augmented Generation
No full textConversational AI (chat) bots are powerful and helpful tools, but are not suited for the unrestricted use in many classrooms: They may hallucinate, easily veer from the topic of instruction, and are vulnerable to malicious prompting. Retrieval-augmented generation (RAG) is a technique that allows educators to constrain chat bots to a specific area of expertise, reducing hallucinations and vulnerability to mis-use. We are working on a low-code solution that enables tech-savvy educators to build such a RAG-based chat bot system themselves, thus retaining full control over the content and behavior of their bot. We present the first version of this system and promising initial feedback from educators and students on its suitability, reliability and flexibility.publishe
