Disentangling decoder capacity and directionality in type constrained knowledge graph link prediction

Résumé fourni par la source

Decoder comparisons in typed knowledge graph (KG) link prediction often confound relation specificity, off-diagonal interactions, and directionality. We separate these factors with four seed-matched decoders applied to GCN, GraphSAGE, and GAT encoders: a shared inner product (IP), a relation-specific diagonal bilinear decoder (RD), a symmetric full bilinear decoder (SB), and an unconstrained directed bilinear decoder (D). The study uses a regulation-grounded coal-mine safety KG with 5,093 nodes and 19,314 canonical directed edges. All 120 decoder–encoder–seed cells use the same ten non-adjacent seeds, type-constrained splits, fixed negatives, and training budget. Two-sided paired AUC tests use pre-specified Holm families, with full filtered MRR and Hits@1/3/10 reported descriptively. RD produced no detectable AUC improvement over IP for any of the three encoders under the matched ten-seed design. SB improved on RD for GraphSAGE ( $$\bf{\Delta}$$ AUC $$\bf{=0.0254}$$ , Holm $$\bf{p=2.15\times10^{-5}}$$ ) and GAT ( $$\bf{\Delta=0.1667}$$ , Holm $$\bf{p=3.24\times10^{-10}}$$ ), but not GCN. Releasing the SB symmetry constraint and enlarging the direction-sensitive score space (D $$-$$ SB) increased AUC for GCN ( $$\bf{\Delta=0.0423}$$ , Holm $$\bf{p=4.77\times10^{-5}}$$ ) and GraphSAGE ( $$\bf{\Delta=0.0116}$$ , Holm $$\bf{p=0.00251}$$ ), whereas the GAT contrast was not detected. Thus, no RD-over-IP improvement was detected, while the contributions of symmetric off-diagonal interaction and of releasing symmetry while enlarging the direction-sensitive score space depend on the encoder. A deterministic semantic hard-negative stress test, all-family filtered ranking, and a frozen-embedding symmetric/skew diagnostic support but do not enter the confirmatory tests. Candidate discovery uses threshold-free top-five within-family ranking. These results support a narrow conclusion: decoder function class matters in this typed transductive setting, but no decoder is universally superior.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.