Human-in-the-(Exploration-)Loop: Visual Pattern-Driven Exploration of Big Datasets, Michael Behrisch
Sprache des Titels:
Englisch
Original Kurzfassung:
Visual Analytics (VA) is the science of analytical reasoning in big and complex datasets facilitated by interactive visual interfaces. Computers are capable of processing enormous amounts of data while humans can creatively pursue their analytical tasks by incorporating their general knowledge. VA systems unite these strengths by allowing the user to interact, understand, and creatively steer the automatic data analysis process.
VA faces, however, challenges like highly specialized expert visualizations, requiring expert model selection, and complex visualization/analysis technique combinations hindering interaction impact. My research pursues a Visual Quality Metrics (VQM) driven approach to overcome these drawbacks. By using quantitative VQMs as visual pattern extractors, analysts can reason over large, complex datasets through exploring interpretable visual patterns in the visualizations.
This talk will demonstrate the overall VQM concept for detecting and making use of meaningful visual patterns with the aim to make data analysis more accessible, effective, efficient, transparent, and reliable. I will show how VQMs and rapid human-in-the-loop interactions can enhance big data exploration by enabling pattern-driven data exploration without relying on specialized visualizations or analysis techniques.
Forschungs-