Entity linking is an essential part of analytical systems for question answering on knowledge graphs (KGQA). The mentioned entity has to be spotted in the text and linked to the correct resource in the knowledge graph (KG). With this paper, we present our approach on entity linking using the abstract meaning representation (AMR) of the question to spot the surface forms of entities. We re-trained AMR models with automatically generated training data. Based on these models, we extract surface forms and map them to an entity dictionary of the desired KG. For the disambiguation process, we evaluated different options and configurations on QALD-9 and LC-QuaD 2.0. The results of the best performing configurations outperform existing entity linking approaches.
CITATION STYLE
Steinmetz, N. (2023). Entity Linking for KGQA Using AMR Graphs. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 13870 LNCS, pp. 122–138). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-33455-9_8
Mendeley helps you to discover research relevant for your work.