Lecture
Nightingale Colloquium presents Spinoza Prize laureate Piek Vossen
- Date
- Tuesday 9 April 2019
- Time
- Explanation
- The seminar is targeted at a broad audience, in particular we invite PhD candidates and supervisors involved in the Data Science Research programme as well as colleagues from LIACS and MI to attend. The seminar is supported by MI and LIACS. The colloquium will be closed with a drink at the FooBar.
- Location
- Snellius
Niels Bohrweg 1
2333 CA Leiden - Room
- 313
Understanding language by machines: from a reading machine to a reference machine
"Identity, reference and perspective play a role when machines need to read large volumes of news but also when robots need to communicate with us about the surrounding world. In this presentation I explain how we use the same Grounded Representation and Perspective (GRaSP) framework to model the semantic parsing of news and the interpretation of communication in a robot brain."
"In the NewsReader project, GRaSP is used to derive both a representation of the changes of the world and the perspective of the source on these changes. I will explain the architecture of the NLP pipeline and as well performance of the modules and the overall system. Detecting events and cross-document event coreference are the main bottle necks for processing large volumes of news. We are currently exploring a data-to-text method to compile data on cross document event coreference which eventually may help solving this bottleneck."
"In the robot-project Leolani, we use an adapted version of GRaSP as a theory-of-mind for dealing with errors, uncertainties and conflicts. In this case, identify, reference and perspective also play a major role as robots and people may interpret the world differently or make different claims. The robot Leolani uses a model of "thoughts" as reflection on the status of her brain, which drives her communication. She also uses a notion of context to be aware of situations, locations and the objects within."
Biography
Piek Vossen is a member of the Royal Netherlands Academy of Sciences (KNAW) since 2017 and recipient of the Spinoza prize (2013). He is a full Professor in Computational Lexicology at the Department of Language, Literature and Communication, VU University Amsterdam. Piek is a leading scholar in language understanding, developed so-called reading machines in the European NewsReader project and recently developing models for communicative robots.
He supervised more than 10 PhDs since he continued his academic career in 2009. He published over 300 peer-reviewed publications in international conferences, journals, books and edited volumes (selection below). In 2013, he received the prestigious Spinoza-prize for his groundbreaking research on wordnets and NewsReader.
He also received the Enlighten-Your-Research prize in 2013 for NewsReader’s challenge to process daily news-streams of millions of articles. He used the Spinoza prize to establish a research group of 20 researchers that work on the semantic processing of text and understanding of language by machines.