Colloquium Details
Scale Diversity
| Author: | Hedda Rahel Schmidtke University of Oregon, Department of Geography |
|---|---|
| Date: | September 01, 2015 |
| Time: | 15:30 |
| Location: | 220 Deschutes |
Abstract
A key phenomenon of the internet is to make space disappear: everything is here, now. Moreover, most computing systems in general are designed to react with no apparent delay, that is within 40ms or 1s reaction time. Computing architectures, whether hardware or software, are designed to meet these requirements, while growing amounts of data have to be processed and stored and hardware is distributed over increasingly larger spaces and numbers of devices. The big data movement as well as technologies used by web giants, such as Google, and the opportunities that are emerging from social media and mobile and ubiquitous computing show that growth in size gives rise to emergent behaviors that were not anticipated, require new technologies, open new opportunities, and give rise to unanticipated dangers. Traditional means for understanding and modeling scale and scalability, such as big-O notation, order of magnitude calculi, or modeling tools, such as UML or Petri Nets, are ill-equipped to handle today's complex, spatially distributed, multi-component systems and big data spaces because of the large number of variables or dimensions involved. Techniques developed in knowledge representation and machine learning are therefore increasingly deployed to monitor and assess systems.
The talk will look at a scenario that is projected to happen over the next 3-5 years as smart phone prices drop below the critical $30 and $20 thresholds: an additional 5 billion users will join the internet from developing countries. I will briefly sketch the current state of the internet in East Africa as I experienced it and demonstrate how aspects of geographic and computational scale impact services in East Africa and how the proposed theory may help to develop products for a global market.
Biography
Hedda R. Schmidtke is assistant professor in the Department of Geography at the University of Oregon. Originally a computer scientist from Germany, she taught in Germany at the University of Hamburg, the Technical University of Braunschweig, and the Karlsruhe Institute of Technology, where she headed the TeCO research group. Other appointments were with Gwangju Institute of Science and Technology in Gwangju, South Korea, and Carnegie Mellon University, which she joined as one of the founding members for its Rwanda campus in Kigali.
Her research and teaching interests revolve around the notions of context and granularity in spatial and spatio-temporal geographic information. On the theoretical side, a research focus is to gain a better understanding of context and granularity in cognition, e.g., their role for resolving vagueness in linguistic expressions when following route instructions, or their role for separating the spatial information contained in maps from thematic content and technical artifacts of generalization. Turning to the application side, a better capturing of context and aspects of context in computing systems is making computing systems smarter, more usable, and suitable to handle increasingly larger amounts of data. Such systems can enable completely new technologies but also raise concerns regarding privacy and reliability that need to be addressed.