These differences are statistically significant, with the t-test showing that MDKG-RL’s improvement over BERT-GNN in MRR (t = 10.00, p = 8.49e-06) and NDCG (t = 6.88, p = 1.25e-04) is highly significant.
← all excerpts
Optimizing document management and retrieval with multimodal transformers and knowledge graphs.
1
—
—