Prof. Dr. Felix Naumann

For our first meeting in this semester, we will discuss organizational issues.

Prof. Dr. Felix Naumann

Abstract: In this meeting we will share the success stories of the students from the past year. In addition, we will have a retrospective of the UCI visit, sharing impressions and learnings. In particular, we will also discuss how the collaboration with UCI can go forward.

Dr. Lisa Ehrlinger

Abstract: High data quality is fundamental for data-based decision making in enterprises and organizations. With the increasing use of computational models to automate or support decision making, the automation of data quality detection and correction becomes more important. In this talk, I will provide an overview on my research on the automation of data quality measurement. By means of industry use cases, I will show how data quality research can be used to solve practical data quality problems. 

Bio: Lisa Ehrlinger is senior researcher at the Software Competence Center Hagenberg (SCCH) in Austria, where she leads the research focus on "Data Management and Data Quality". She received her diploma in computer science and her PhD on "Automated Continuous Data Quality Measurement", both from the Johannes Kepler University (JKU) Linz. Her research interests and publications cover the topics data quality, semantic technology, knowledge graphs, metadata management, and information integration. L. Ehrlinger is also member of the Global Editorial Board of the CDO Magazine and lecturer at JKU on databases, information systems, and ontology engineering.

Dr. Timo Kötzing

Abstract: During coursework we all learned how to deal with the details of our scientific discipline: whether it is developing software or proofs, we have all the tools we need. In this talk we cover some strategies and tricks on anything in between these phases of dealing with the details -- the dealing with the bigger picture. I finished a project, what should I do next? We have visiting scholars, how can we find a common working direction? The results are in, but they don't show what we thought they would show; how do we go from here? Furthermore, many of the strategies can be applied to other general questions like "What should I propose in a grant proposal?" or "How do I present my results in a conference talk?"

Katerina Niklanovits

Abstract: Partitioning a graph into connected components is a fundamental task in graph algorithms, and those connected partitions occur as desirable structures in areas such as image processing, network decomposition and robotics. In 1976, Gyori and Lovasz independently proved that any k-connected graph G can be partitioned into k disjoint connected components each of them containing a given vertex and having a desired size. Through the last years, polynomial time algorithms have been provided for the cases where k=2,3,4, but the cases where k is greater than 4 remain open. In this talk a polynomial time algorithm for Gyori Lovasz theorem is presented, for any k when G is a chordal graph. This result is joint work with Katrin Casel, Tobias Friedrich,  Davis Issac and Ziena Zeif.

Gerardo Vitagliano

Abstract: Powerpoint or latex? Animations or transitions?  Live or recorded? In this session we will go through how to present research work effectively in an oral presentation.
We will cover the most important dimensions to think about when preparing and giving presentations, such as timings, story-telling, and slides design. 
The session will be an interactive discussion, so be prepared to share your thoughts, experiences (good or bad) and ideas!

Prof. Dr. Gerard de Melo

Abstract: In science, it can often take months or even years of hard work to 
obtain novel research results. However, just doing good research is not 
enough. If no one hears about this research, one could argue that you 
haven't yet made a genuine contribution to science. In a previous talk, 
I focused on promoting your research via social media and other 
marketing channels. This talk will present a broader range of measures 
that can help in informing your community about your research. For 
example, we will discuss some questions that arise when deciding how to 
publish your paper, as well ways of ensuring a full-fledged public 
release of your results that others can more easily digest and build upon.

Prof. Dr. Ralf Herbrich

Abstract: writing (well) is one of the key skills for any researcher; scientific discourse is about discovering new knowledge and this is happening through conversations and communications between researchers. Scientific documentation is a key tool for communication and conversation. In this talk, I will give an overview about the typical kinds of scientific documentation and share tips, tricks and tools for writing well.

Tarek Galal

Abstract: Digital COVID Certificates are the first widely deployed end-user cryptographic certificates. Their verification constituted a challenge to service provides that need to check the validity, due to the complexities and possibly evolving nature of required policies (e.g., 2G, 3G, in the COVID context). To address this, the European Commission proposed a specification for outsourcing the verification process to a Validator Service who executes the process and informs service providers of the result. Outsourcing verification is useful not only to COVID certificates, but generally to end-user certificates, as it is a convenient solution for service providers and can improve security and privacy - when done properly - in online scenarios. 

Therefore, we formally model outsourced validation services and define the desired security and privacy properties of such services.  We analyze the EU specification in our model and show that it only meets a subset of those goals. Our analysis further shows that the protocol can be simplified and still maintain the same (weak) level of security. Finally, we propose a new construction for privacy-preserving credential validation that provably satisfies all desired goals.

Prof. Dr. Tilmann Rabl

Abstract: In this presentation, we will discuss the intricacies of receiving and giving reviews. We'll start by looking at the review process in DB conferences. Then discuss types of reviews one can receive and how to read them. In the last part, we briefly discuss how to write reviews.

Wang Yue

Abstract: In past decades, more and more devices are part of the Internet-of-Things (IoT). These devices are widely used in industry as well as in research and are involved in many applications. The massive amount of devices from these applications are connected in many large decentralized networks. Many unbounded data streams are produced from decentralized networks and generate data continuously. To process data in a timely fashion, current stream processing engines (SPEs) collect all data streams to their data centers. They split data streams into windows and perform window aggregations to calculate results. That leads to high network overhead and makes data centers bottlenecks, as all data is computed centrally. State-of-the-art solutions push down partial window aggregations to machines that are near data streams. Data centers process partial results instead of raw data. However, these solutions are limited to simple queries that have decomposable functions and time-based window measures. They process data centrally when windows have non-decomposable functions

We present Dema, a decentralized window aggregation technique that is designed for non-decomposable functions. Dema can off-load window aggregations to the data sources and only need to collect partial results and partial data to its center. Furthermore, Dema is applied for windows that have time-based and count-based measures. In our evaluation, we demonstrate that Dema outperforms state-of-the-art solutions by orders of magnitude when processing windows with non-decomposable functions and any window measures. Dema can reduce 99% of network traffic and significantly improve the throughput of data centers.

Prof. Dr. Bernhard Renard

Abstract: Whether it is computational benchmarks or actual data collection, designed experiments are a key way of generating scientific evidence and are at the heart of many data-centric PhD theses, Within this talk, I will summarize a checklist from Angela Dean’s standard textbook on experimental design and combine with my own experience of aspects to watch out for. This will stretch from goal definition, via variance factors, lurking variables and randomization schemes to sample size considerations. Participants are encouraged to share their own experience and to compare standards across fields.