Multilingual Knowledge Graphs Question Answering

Nikit Srivastava
Tatiana Moteu Ngoli
Research Theme
R1 Human agency
Knowledge Graphs Multilingual Language Models

This project is centered on devising techniques and methodologies to improve the state of Multilingual Question Answering over Knowledge Graphs. Our objective is to work towards developing systems that are able to comprehend and generate SPARQL from a diverse set of languages natively. Given the nature of the task involving transformation of natural language queries into SPARQL, we aim to leverage the capabilities of (large) Language Models (LMs) for this purpose. Therefore, LMs form a fundamental aspect of this research as we extend our efforts to improve the performance on the multilingual variants.