Deep learning over relational databases has led to increasingly effective prediction models, motivating a growing need to explain their behavior. We study explanation methods that assign importance scores to individual database attributes, focusing on a fundamental question: to what extent does the choice of importance-attribution method matter? We develop a methodology to systematically compare importance-attribution methods and analyze their similarities and differences. In particular, we study two aspects of the explanations: which attributes they deem important, captured by differences in attribute rankings, and how importance scores vary across attributes, captured by differences in score distributions. We further investigate whether these differences are reflected in the conventional measure of explanation quality, fidelity. We apply our methodology in an extensive empirical study of four methods: Monte Carlo Shapley, Kernel SHAP, single-attribute impact, and mask learning. We experiment with multiple datasets and tasks using two RDL architectures: heterogeneous GraphSAGE and Relational Graph Transformers. Our experiments show that explanation methods behave differently depending on the prediction type: regression or classification. In regression, most methods exhibit high agreement in attribute rankings, whereas classification yields considerably less consensus. Notably, while mask learning frequently diverges in its importance distributions, Shapley-based methods and single-attribute impact often converge to similar explanations. This suggests that, in many settings, single-attribute impact can replace Shapley at substantially lower computational cost. Overall, our study provides methodological tools for comparing importance-attribution methods, and empirical evidence that their choice has important consequences for interpreting relational deep learning models.
Figures & tables
Figure 1. A representative empirical convergence curve (shown for HGS on the rel-trial dataset for the study-adverse task). The plot tracks the absolute difference in estimated scores ( ∣Δshapley_value∣ ) between consecutive permutation checkpoints.
Figure 2. Pairwise explainer similarity for regression, showing alignment on different prefixes ( k ) for HGS and RelGT.
Figure 3. Pairwise explainer similarity for classification, showing alignment on different prefixes ( k ) for HGS and RelGT.
Figure 4. Score distribution (PDF). Dots show the percentage of attributes within the normalized weight range.
Figure 5. Normalized score scatter grid. Attributes along the x-axis are sorted by collective importance.
Figure 6. Consensus min-max normalized score grids across all tasks (top-15 evaluation, displaying top 10 consensus attributes). Cells display individual ranks ( # ) and bar widths represent normalized scores. Data type are in brackets: [N] Numerical, [C] Categorical, [D] Date, and [T] Text. Alphabetical tie-breaking may affect the exact top- k choices (See Section 5.2 ).
Figure 7. Explainer fidelity in regression (top 2 rows) and classification (bottom 2 rows) tasks, over HGS (left) and RelGT (right). Bars represent Fidelity+ (necessity, higher is better) and Fidelity− (sufficiency, lower is better) across varying top- k attribute subsets. The maximum y-axis value for each subplot is annotated in its top-left corner.
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 8. Empirical convergence curves for the Monte Carlo Shapley approximation on the rel-f1 and rel-event datasets.
Figure 9. Empirical convergence curves for the Monte Carlo Shapley approximation on the rel-trial and rel-arxiv datasets.
Figure 10. Empirical convergence curves for the Kernel SHAP approximation on the rel-f1 and rel-event datasets.
Figure 11. Empirical convergence curves for the Kernel SHAP approximation on the rel-trial and rel-arxiv datasets.
Machine learning pipelines commonly flatten relational data into single-table representations, discarding structural constraints. Widely used Shapley value-based feature attributions then rely on feature independence, evaluating the model on combinations that could never arise in the underlying data, producing misleading explanations. We propose RelShap, a framework that incorporates relational constraints and data provenance into Shapley value computation, restricting both background data and coalition evaluation to relationally valid configurations. The framework is estimator-agnostic and composes with Kernel SHAP, Monte Carlo, and Leverage SHAP without altering their sampling or weighting properties. Functional dependencies further induce equivalence classes over feature coalitions, which RelShap exploits to reduce runtime without changing Shapley values; we provide a combinatorial characterization of the expected speedup. Experiments across multiple datasets, models, and estimators show that RelShap produces explanations that are more faithful to the data-generating process, correctly identifying the dominant feature in controlled settings where existing methods, including Conditional SHAP and ManifoldShap, do not. Our code is available at: https://github.com/duneag2/relshap.
Seungeun Lee, Joao Fonseca, Julia Stoyanovich
New York University, New York, USA · INESC-ID, Lisbon, Portugal
Relational databases (RDBs) are the primary data infrastructure in many enterprises, yet recent deep learning methods designed for RDBs have been evaluated under inconsistent experimental protocols, making fair comparison difficult. We present one of the first systematic benchmarking studies of recently released deep learning methods for RDBs, evaluating them across five relational databases, with one classification and one regression task for each. We refactor all deep RDB models to allow the full range of experimental procedures to be applied consistently across all methods. Our findings indicate that the relational transformer (RT) approach delivers the strongest overall performance on both classification and regression tasks compared to the state-of-the-art graph-based modeling and learning of RDBs. Even for single-table learning tasks, deep learning methods designed for RDBs outperform the leading tabular foundation model, TabPFN 2.5. Extending learning from a single table (hop = 0) to multiple tables (hop = 1, 2) by connecting neighboring tables in relational databases enhances performance, but the additional benefit from higher hops diminishes as computational overhead grows. Deep RDB learning methods have the potential to challenge state-of-the-art tabular foundation models, especially on large-scale enterprise data. The source code for this benchmarking study is publicly available.
Kazi F. Akhter, Bharath Ajendla, Manar D. Samad
Department of Computer Science Tennessee State University Nashville, TN, USA · SAP AI Research and Innovation SAP Labs Palo Alto, CA, USA
Visual attribution methods typically explain predictions using pixels, superpixels, or regular patches. These representations can localize important regions, but provide limited information about their structure. We introduce MorphoSHAP, a model-agnostic post-hoc method that instead uses morphological shapes as the players of a Shapley attribution game. Using the Tree of Shapes, each shape is described by its scale, geometry, and signed contribution, providing explanations of where the evidence lies, what type of structure carries it, and how strongly it affects the prediction. This shared morphological vocabulary enables spatial, textual, and global class-level explanations beyond image-specific heatmaps. To the best of our knowledge, MorphoSHAP is the first SHAP-based image attribution framework to combine these different forms of explanation. Across five diverse datasets and three architectures, MorphoSHAP achieves strong insertion/deletion performance and outperforms competing attribution methods on several benchmarks. Finally, a user study shows that MorphoSHAP provides explanations that are easy to use and are preferred over standard attribution baselines.