Knowledge Graphs meet Language Models
Knowledge Graphs represent information from real-world entities and their relationships in a graph structure. Their construction, representation and completion have been an active research topics in the last few years.
On the other hand, information have been also represented through natural language in vast amounts of documents throughout the history of humanity and recently large language models have been trained to embed this textual information in dense vector spaces.
In this seminar, we are going to study the basic concepts of knowledge graphs (creation and completion) and language model training, and then we will concentrate on recent research that have tried to combine both approaches.
Supervisors: Felix Naumann, Nitisha Jain and Alejandro Sierra-Múnera
The introductory, first session is open to all and will take place on Monday, April 25th from 13:30 to 14:30 in F.E-06.
Learning Goals
Students will learn to...
- read and understand scientific publications
- analyze and summarize research contributions
Schedule
Time: 13:30- 15:00
Location: Campus 2, F.E-06 (Changed to F-2.10)
| July 21 | Final Poster session (14:00) |
(still updated, please not the changes)
Grading
- Paper presentation 30%
- Final poster presentation 70%
Prerequisites
- Basics of machine learning and neural networks
Resources
- Gerhard Weikum, Xin Luna Dong, Simon Razniewski and Fabian Suchanek (2021), "Machine Knowledge: Creation and Curation of Comprehensive Knowledge Bases", Foundations and Trends® in Databases: Vol. 10: No. 2-4, pp 108-490. http://dx.doi.org/10.1561/1900000064
- Dan Jurafsky and James H. Martin, "Speech and Language Processing" (3rd ed. draft) https://web.stanford.edu/~jurafsky/slp3/