Heterogeneous Graph Data Valuation: A Shapley Value-based Approach
Bibliographic record
Abstract
With the widespread application of graph neural networks on heterogeneous graph data, accurately assessing the contribution of each node to model performance has become increasingly important. However, most existing graph data valuation methods are designed for homogeneous graphs and fail to fully account for the unique roles of different node types in heterogeneous graphs. This paper proposes a heterogeneous graph data valuation method based on an improved Shapley value, which effectively reduces computational complexity and enhances valuation accuracy by constructing local subgraphs and incorporating node-type information. Specifically, this paper first designs a subgraph construction strategy that considers node types, capturing local structural information by retaining key neighbor nodes. Second, it introduces an improved Shapley value calculation method, leveraging hierarchical sampling and type-aware importance evaluation to improve computational efficiency. Finally, a node-type-based validation mechanism is proposed, which verifies the valuation results by observing the impact of progressively removing nodes of different types on model performance. Experiments on the AIFB dataset demonstrate that the proposed method not only accurately identifies nodes significantly influencing node classification tasks but also reveals unique contribution patterns of different node types in heterogeneous graphs. The results highlight that publication-type nodes have the most significant impact on classification performance in academic knowledge graphs, providing critical guidance for the filtering and simplification of heterogeneous graph data.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.023 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.007 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".