The goal of this project is to study and develop algorithms to improve the way we interact with software based on extenshive data analytics. Our focus lies in extracting and aggregating information from large data sources to drive personalized interactive software systems, such as intelligent tutoring systems that improve the way students learn.
Writing and Language
Specific learning disabilities describe problems in learning specific tasks which can not be explained with general cognitive difficulties, sensory acuity or general motivation. A common learning disability is developmental dyslexia. Children and adults affected by developmental dyslexia typically are slow readers, display limited reading comprehension, and have inconsistent orthography speed and accuracy problems as well as difficulties in segmenting and manipulating phonemes in words. Approximately 10% of the population suffers from dyslexia. We develop a novel multi-sensory training software for dyslexia. We use computer graphics to circumvent the difficulties and to illustrate the learning content. The software program hence consist of a multimedia framework. Additionally, we use student modelling to have the software adapt to the user's skills and support his/her learning process in the best way possible.
In today's society the possession of mathematical skills is crucial as numerical cognition and calculations are ubiquitous in everyday life. However, 3-6% of the population in English- and German-speaking countries suffer from developmental dyscalculia (DD), a specific learning disability affecting the acquisition of arithmetic skills. People affected by dyscalculia have difficulties performing mathematical tasks. Sometimes simple tasks such as estimating the change in a shop are hard to solve. Causes for dyslcalculia are considered to be difficulties in mapping amounts to numbers and figures as well as an under-developped internal number line. At CGL, we develop a novel multi-sensory training software for dyscalculia. We use computer graphics to circumvent the difficulties and to illustrate the learning content. We employ student models to personalize the learning experience to increase the efficiency of learning.
Personalizing interactive systemsData analytics has the potential to fundamentaly change the way we interact with computer-based systems. To personalize interactive experiences a detailed understanding of the user is necessary. For this we are researching data-driven algorithms that allow the construction of detailed user models tracking user properties such as knowledge acquisition or engagement across different domains.
Scheduling and PreferencesToday calendar tools are widely adopted for both personal and commercial purposes. Nevertheless most decision-making responsibilities, as well as reasoning about scheduling preferences and constraints, are still left to the user. This project aims at exploring the benefit of modern Machine Learning and Cognitive Computing methods to upgrade a calendaring tool to a personalized virtual assistant with autonomous or semi-autonomous decision-making capabilities.