SEMANTiCS 2019 Research & Innovation Chair Philippe Cudre-Mauroux is a full Professor at the University of Fribourg. The focal areas of his research are Exascale Information Management for AI, Big Data, Scientific Data and Linked Data. In this Interview, Philippe talks about the approaches he takes at the eXascale Infolab, current projects as well as expectations and advice regarding research and innovation papers for SEMANTiCS 2019.
You lead the eXascale Infolab at the University of Fribourg. What are your primary areas of interest? Which topics are you currently working on?
The eXascale Infolab’s mission is to design, build and deploy new software infrastructures for Big Data and AI. Database researchers spent decades exploring the relational model. In my lab, we look at other data types such as graphs, vectors, or arrays, and come up with new, efficient ways to store and process them on clusters of commodity machines. Some of the new problems we are currently tackling include designing systems for new hardware (e.g., how to take advantage of programmable switches to aid query processing) or for new Machine Learning applications (e.g., more efficient systems to create text or graph embeddings).
You lead a lot of really exciting projects at eXascale Infolab. One of your latest projects is sponsored by the European Research Council (ERC), an agency supporting frontier research. Can you tell us more about it?
My ERC project GraphInt, which is currently running, is devoted to semi-automatically creating Knowledge Graphs from unstructured or semi-structured content. Contrary to current approaches, we aim at creating and managing incredibly detailed Knowledge Graphs capturing most of the information from the original source in a structured manner. My team and I have pioneered a number of exciting techniques in that sense, including new link prediction approaches using active learning, or human-in-the-loop probabilistic systems to debug noisy Knowledge Graphs (see for instance our upcoming papers at The Web Conf WWW2019 on these topics).
Browsing through your work I was fascinated by the MEM0R1ES, a project addressing the design of information systems on basis of the idiosyncrasies of human memories. Could you also elaborate a bit on this project - what is the objective and what is the outcome?
End-users routinely produce massive amounts of digital information. Once created, this information is often stored locally on a particular device or remotely on a proprietary platform. Related pieces of content are hence inevitably scattered across a collection of digital silos that prevent the sharing, the cross-pollination, and the long- term archival of the various pieces of information. MEM0R1ES is a platform tackling this issue, enabling digital denizens to reclaim their digital life and to preserve and integrate all the data they daily produce in a highly-available, secure, scalable, and semantically rich digital store.
At SEMANTiCS 2019 you will be chairing the Research and Innovation Track. What do you expect by reviewing the incoming submissions? Please share some insights about your perspective on research and innovation projects that currently are on your radar.
SEMANTiCS is the leading conference bridging the gap between researchers and practitioners in semantic technologies and AI. In that sense, I expect a number of exciting contributions promoting the wide adoption of semantics and graph technologies. Applications mixing Knowledge Graphs and AI might be particularly hot this year. Knowledge Graphs have the potential to bring AI models to the next level, by providing different training data or by allowing humans to express constraints on a particular model.
Any last words to those who haven’t submitted yet, researchers facing constraints during the work on their projects or those who still have a lot on their ToDo-Lists before the calls close in April
There’s never been a better time to share your thoughts on semantics and Knowledge Graphs, do submit! SEMANTiCS is welcoming both new ideas and new industrial applications in this domain. My colleagues at the Program Board and I compiling a thrilling program mixing top academics and industry players and it would be a shame to miss it.
The annual SEMANTiCS conference is the meeting place for professionals who make semantic computing work, and understand its benefits and know its limitations. Every year, SEMANTiCS attracts information managers, IT-architects, software engineers, and researchers, from organisations ranging from NPOs, universities, public administrations to the largest companies in the world. http://www.semantics.cc