|
On Language Models and Knowledge Bases | title | On Language Models and Knowledge Bases |
|---|
| start_date | 2025/11/21 |
|---|
| schedule | 11h |
|---|
| online | no |
|---|
| location_info | Paul Erdős (A115) room & online |
|---|
| summary | Language Models have brought major breakthroughs in naturallanguage processing. Notwithstanding this success, I will show that certain applications still need symbolic representations. I will then show how different methods (language models and others) can be harnessed to build such symbolic representations in the form of knowledge bases. I will highlight several challenges in this endeavor, from finding good embeddings to improving entity linking and dealing with fallacies and textual entailment. I will also discuss how language models can be evaluated along several dimensions. Finally, I will talk about the knowledge bases themselves, most notably our YAGO project. I will present our work on detecting and alleviating incompleteness in knowledge bases, on querying the data, on using the data for the digital humanities, and on reasoning on beliefs. |
|---|
| responsibles | Bawden |
|---|
Workflow history| from state (1) | to state | comment | date |
|---|
| submitted | published | | 2025/11/10 09:48 UTC |
| |
|