The paper presents a neuro-symbolic framework, PSYCHIC, developed for the Scholarly Question Answering over Linked Data (Scholarly QALD) challenge at The International Semantic Web Conference 2023. This challenge focuses on question answering (QA) over knowledge graphs (KGs). The paper’s authors propose a system that achieved a F1 score of 00.18% on question answering and came in third place for entity linking (EL) with a score of 71.00%. The system integrates large language models with knowledge graphs to mitigate the shortcomings of large language models, such as their propensity to disseminate incorrect knowledge.
Publication date: 19 Oct 2023
Project Page: https://kgqa.github.io/scholarly-QALD-challenge/2023/
Paper: https://arxiv.org/pdf/2310.12638
