ABSTRACT
Like most curricula in the humanities and social sciences, the curriculum of pre-service teacher training in educational sciences often includes time-consuming reading and writing tasks, which require high quality support and feedback in a timely manner. A well-known way to provide this support to students is one-to-one mentoring. However, limited time and resources in the German university context require to effectively scale the benefits of individual feedback. The use of scalable technologies to support learning processes seems to be promising, but its development usually requires a deep technical understanding. With an interdisciplinary approach, this contribution investigates how personal mentoring can be made available to as many students as possible, taking into account the didactic, organizational and technical frameworks at universities. We describe the development and implementation process of two chatbots that both aim to support students of educational sciences in their self-study of the seminar topics and literature. The chatbots were used by over 700 students during the course of 1 year and our evaluations show promising results that bear the potential to improve the availability of digital mentoring support for all students.
ABSTRACT
The accuracy of fingerprinting-based indoor localization correlates with the quality and up-to-dateness of collected training data. Perpetual crowdsourced data collection reduces manual labeling effort and provides a fresh data base. However, the decentralized collection comes with the cost of heterogeneous data that causes performance degradation. In settings with imperfect data, area localization can provide higher positioning guarantees than exact position estimation. Existing area localization solutions employ a static segmentation into areas that is independent of the available training data. This approach is not applicable for crowdsoucred data collection, which features an unbalanced spatial training data distribution that evolves over time. A segmentation is required that utilizes the existing training data distribution and adapts once new data is accumulated. We propose an algorithm for data-aware floor plan segmentation and a selection metric that balances expressiveness (information gain) and performance (correctly classified examples) of area classifiers. We utilize supervised machine learning, in particular, deep learning, to train the area classifiers. We demonstrate how to regularly provide an area localization model that adapts its prediction space to the accumulating training data. The resulting models are shown to provide higher reliability compared to models that pinpoint the exact position.
