Dr. Yulia Zinova

Topic: Word Embeddings and Morphology - Do language models learns what linguists are describing?

Language models seem to "understand" natural language to some extent - some well-known systems can answer questions, provide summaries, write abstracts, etc. At the same time, these systems dramatically fail at some tasks a preschool child can easily complete. As a linguist, I'm interested not only in improving the performance of language models in the downstream tasks, but also in understanding what these models learn and how does it relate to the assumptions and findings in theoretical linguistics. In this talk I will present some of the ongoing research on morphology with static embeddings. I will talk about what kind of information gets implicitly learned, which limitations the models face and how linguistic knowledge can be used to boost performance of language models, especially for languages with less resources and/or rich morphology.

Dr. Kilian Evang

Topic: Semantic Roles for Semantic Parsing - Lessons Learned and Future Directions

In this talk I report on efforts to annotate a parallel Role and Reference Grammar treebank (RRGparbank) with semantic roles for verbs. I discuss the annotation scheme developed, the annotation process, and semantic parsing results. Finally, I highlight some problems with existing annotation schemes based on VerbNet, PropBank, FrameNet, and VerbAtlas, and sketch a new scheme that solves some of these problems.

Prof. Dr. Kevin Tang

Topic: Computational models in Parkinson's Disease research

Speech can be used as a non-invasive biomarker to capture fine changes in speech patterns in normal populations and individuals diagnosed with neuromotor disorders, such as Parkinson's Disease (PD). In this talk, I will demonstrate how computational models that are linguistically-informed can quantify acoustic and syntactic variations in PD patients.

Dr. Nurul Lubis

Topic: Dialogue evaluation via Offline Reinforcement Learning and Emotion Prediction

Task-oriented dialogue systems aim to fulfill user goals, such as booking hotels or searching for restaurants, through natural language interactions. They are ideally evaluated through interaction with human users. However, this is unattainable to do at every iteration of the development phase due to time and financial constraints. Therefore, researchers resort to static evaluation on dialogue corpora. Although they are more practical and easily reproducible, they do not fully reflect real performance of dialogue systems. Can we devise an evaluation that keeps the best of both worlds? In this talk I explore the usage of offline reinforcement learning and emotion prediction for dialogue evaluation that is practical, reliable, and strongly correlated with human judgements.

Apl.-Prof. Dr. Wiebke Petersen

Topic: In progress

More details will be announced soon, please check back later!

Accommodation

Foto Ⓒ Jugendherberge Düsseldorf
Düsseldorfer Str. 1,
40545 Düsseldorf
From here, you can reach the venue by bus in only 20 minutes. In the unlikely case that you would like to skip some talks during the conference, you may take a stroll at the Rhine or just go to town and enjoy Düsseldorf, which is right on the other side of the river.

Venue

Foto Ⓒ Heinrich-Heine-Universität-Düsseldorf
Our beloved Alma Mater in her entire brutalist architecture glory. The campus of the HHU stretches nearly 4km long, but the conference will take place inside of the newly refurbished building 23.21, which houses the linguistics department, the PC-pools, and the cafeteria of the humanities department. The lecture halls, rooms, and floors that are reserved for keynote, and student talks, poster presentations etc. will be announced soon.

Special Venue

Foto Ⓒ Sigurd Steinprinz.
This "House of the University" was built in 1897 and was originally not operated by the HHU and its friends. Since 2013, It serves as an oasis of knowledge in the city center and hosts public events, conferences and exhibitions. We managed to book a day or two here, so the conference can get a proper opening with some grand, and official air around it.