- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper studies the mismatch between observed graph structure and textual semantics in text-attributed graphs. The authors propose SEGIA, which first uses an LLM to augment node text with predicted classes and explanations, then applies a variational encoder to learn latent semantic representations, posterior uncertainty, and soft block memberships. These representations are used to construct a semantic graph through an SBM-inspired edge scorer that combines block compatibility, node activity, membership ambiguity, posterior dispersion, and latent distance. The observed and semantic graphs are then processed by separate GCN branches, whose predictions are aligned using an optimal transport objective. Experiments on five citation-network benchmarks show that SEGIA consistently outperforms the baselines considered in the paper.
Critical strengths
- The paper identifies a practically relevant structural–semantic mismatch in text-attributed graphs and provides a coherent framework that preserves the observed topology while introducing an additional semantic graph inferred from textual information.
- The overall method is clearly presented and reasonably well organized. The proposed semantic graph construction combines multiple sources of information, including soft block compatibility, node activity, semantic proximity, and posterior dispersion.
- The results are strong within the evaluated setting. SEGIA achieves the best reported accuracy among the compared methods on all five benchmarks, and the paper provides ablations covering the semantic graph, observed topology, OT alignment, uncertainty, SBM scoring, and LLM expansion.
Critical weaknesses
- The methodological novelty seems relatively incremental. Most components of SEGIA are based on established techniques, and the main contribution appears to lie in their combination rather than a fundamentally new approach.
- Several closely related methods discussed in the paper, especially TAPE, GAugLLM, and LLaTA, are missing from the main comparison. This makes it difficult to judge SEGIA against the most relevant alternatives.
- The current ablations do not isolate the individual components of the proposed edge scorer. Replacing the full scorer with cosine similarity cannot determine which terms are responsible for the improvement.
- The evaluation is limited to citation networks, which leaves the broader applicability to other types of text-attributed graphs unclear.
Questions for authors
- Could the authors provide statistics or qualitative analysis of the learned semantic graph, such as the fraction of candidate pairs not present in the observed graph, the distribution of weights assigned to observed versus newly introduced candidate pairs, or representative examples of highly and weakly weighted connections?
- How does SEGIA behave as the mismatch between topology and semantics increases? For example, could the authors evaluate the model under controlled edge deletion or addition?
- What is the computational cost of SEGIA, including LLM preprocessing and semantic graph construction, compared with the closest LLM-based TAG methods? In particular, how does the cost on OGBN-ArXiv?
3 comments
W1: SEGIA's contribution lies in how textual semantics and uncertainty jointly inform graph inference and how the inferred graph is integrated with the observed topology. Specifically, variational encoding provides latent representations and uncertainty estimates, which the SBM-inspired scorer uses to assign continuous weights to candidate relations. Separate graph branches retain the observed and inferred relations, while prediction alignment coordinates the two views. The semantic-graph analyses and individual scorer ablations provide evidence for these design choices.
W2: Following your suggestion, we evaluated TAPE and LOGIN using the same LLM-augmented text inputs as SEGIA (Table 7). Reproducing GAugLLM requires additional auxiliary LLM outputs and configurations, and its default Mixtral 8×7B model exceeds our computational capacity. LLaTA's code is not publicly available. We therefore included LOGIN as another closely related baseline that uses LLM feedback to guide graph refinement.
Table 7. Test accuracy (%) of closely related methods.
| Model | Cora | Citeseer | PubMed | ArXiv-2023 | OGBN-ArXiv |
|---|---|---|---|---|---|
| TAPE | 87.64±1.10 | 74.51±2.37 | 93.50±0.54 | 78.13±0.23 | 76.27±0.38 |
| LOGIN | 86.86±2.10 | 72.82±2.68 | 82.26±0.60 | 74.64±0.49 | 72.44±0.24 |
| SEGIA | 91.00±0.71 | 81.32±1.03 | 90.45±0.22 | 78.20±0.26 | 76.54±0.06 |
SEGIA achieves the highest accuracy on the four datasets other than PubMed, while TAPE achieves higher accuracy on PubMed (93.50% vs. 90.45%). One possible explanation lies in their different text representation strategies. PubMed contains domain-specific biomedical terminology, for which TAPE's task-specific LM fine-tuning may provide more adaptive representations. In contrast, SEGIA employs a frozen text encoder, potentially limiting its adaptation to specialized domains. This suggests a trade-off between task-specific representation adaptation and computational efficiency. Notably, across five runs on PubMed, SEGIA requires only 3,073 s, compared with 11,308 s for TAPE, achieving approximately 3.68× faster execution.
W3: Following your suggestion, we conducted ablations that individually remove block compatibility, membership entropy, node activity, posterior variance, or latent distance from the edge scorer. As shown in Table 8, removing each scoring term reduces mean accuracy on all five datasets, supporting the contribution of each term to the full scorer.
Table 8. Test accuracy (%) with individual scorer components removed.
| Model variant | Cora | Citeseer | PubMed | ArXiv-2023 | OGBN-ArXiv |
|---|---|---|---|---|---|
| Without block compatibility | 89.41±1.29 | 76.05±2.22 | 89.34±0.15 | 78.07±0.11 | 76.42±0.02 |
| Without membership entropy | 89.70±1.42 | 75.96±1.95 | 89.63±0.52 | 77.97±0.06 | 76.40±0.04 |
| Without node activity | 89.52±1.81 | 76.68±1.89 | 89.55±0.54 | 77.99±0.21 | 76.32±0.06 |
| Without posterior variance | 89.70±1.72 | 76.46±1.48 | 89.62±0.43 | 78.02±0.15 | 76.36±0.11 |
| Without latent distance | 89.41±1.83 | 75.80±1.90 | 89.45±0.41 | 77.97±0.33 | 76.30±0.03 |
| Full scorer | 91.00±0.71 | 81.32±1.03 | 90.45±0.22 | 78.20±0.26 | 76.54±0.06 |
W4: Following your suggestion, we extended the evaluation to Computers and Products, two product co-purchase graphs. As shown in Table 9, SEGIA achieves the highest mean accuracy on both datasets, supporting its applicability beyond citation networks.
Table 9. Test accuracy (%) on Computers and Products using augmented text embeddings.
| Model | Computers | Products |
|---|---|---|
| MLP | 69.13±0.36 | 80.02±0.41 |
| GCN | 86.86±0.28 | 83.81±0.47 |
| GAT | 87.32±0.22 | 82.83±0.43 |
| DeloopSGNN | 87.93±0.14 | 83.83±0.34 |
| AdaSpec | 87.89±0.28 | 81.77±0.60 |
| RADE | 89.03±0.15 | 85.18±0.45 |
| AD-GCN | 87.03±0.20 | 86.07±0.39 |
| CGT | 86.60±0.28 | 83.70±0.44 |
| SEGIA | 89.09±0.40 | 87.98±0.31 |
Thank you for the detailed responses and additional experiments. I appreciate the authors’ efforts to address my concerns. These additions have substantially strengthened the result. I would like to increase my rating to 3. Thank you again for the responses and discussion.
Q1: 1) Following your suggestion, we measured the proportion of candidate edges absent from the observed graph: 76.60%, 65.68%, 77.13%, 91.70%, and 24.80% for Cora, Citeseer, PubMed, ArXiv-2023, and OGBN-ArXiv, respectively. Table 1 reports the weight distributions for observed and added edges, and Table 2 reports label homophily.
Table 1. Weight distributions of observed and added edges.
Weights are the final symmetrized values before degree normalization.
Table 2. Label homophily (%) of observed and semantic graphs and high-weight added edges.
As shown in Table 2, the weighted semantic graph and high-weight added edges have higher label homophily than the observed graph on all five datasets. These results indicate that the learned weights emphasize class-consistent relations.
2) We also inspected representative high- and low-weight connections in Cora. Two papers on reinforcement learning for robotics (Cora IDs 55968 and 1152663) have no observed citation edge but receive a semantic edge weight of 1.8149. By contrast, an observed edge between papers on learning theory and recurrent neural networks (IDs 12350 and 16451) receives a weight of 0.0502. These examples illustrate how the scorer assigns high weights to semantically related pairs without observed edges and low weights to less related observed connections.
Q2: Following your suggestion, we evaluated SEGIA under 30% random edge addition and deletion to assess robustness to spurious and missing connections, respectively. As shown in Tables 3 and 4, SEGIA achieves the highest accuracy among the evaluated methods on all five datasets under both perturbations.
Table 3. Test accuracy (%) under 30% random edge addition.
Table 4. Test accuracy (%) under 30% random edge deletion.
Q3: Following your suggestion, we report one-time text preprocessing separately from training and evaluation (Tables 5 and 6). On OGBN-ArXiv, all three methods share the same preprocessed inputs, with a common one-time preprocessing cost of 27,778.8 s. Subsequent execution, including semantic graph construction and scoring, takes 919.25±3.94 s for SEGIA, 26,238.39±19.78 s for TAPE, and 1,672.99±15.01 s for LOGIN. SEGIA therefore has the lowest total runtime among the three methods.
Table 5. One-time text preprocessing time (min).
Table 6. Training and evaluation time per run (s; mean ± sample standard deviation).