Small and Medium Enterprises (SMEs) constitute 99.9% of U.S. businesses and generate 44% of economic activity, yet systematically identifying high-potential SMEs remains an open challenge. We introduce SME-HGT, a Heterogeneous Graph Transformer framework that predicts which SBIR Phase I awardees will advance to Phase II funding using exclusively public data. We construct a heterogeneous graph with 32,268 company nodes, 124 research topic nodes, and 13 government agency nodes connected by approximately 99,000 edges across three semantic relation types. SME-HGT achieves an AUPRC of 0.621 0.003 on a temporally-split test set, outperforming an MLP baseline (0.590 0.002) and R-GCN (0.608 0.013) across five random seeds. At a screening depth of 100 companies, SME-HGT attains 89.6% precision with a 2.14 lift over random selection. Our temporal evaluation protocol prevents information leakage, and our reliance on public data ensures reproducibility. These results demonstrate that relational structure among firms, research topics, and funding agencies provides meaningful signal for SME potential assessment, with implications for policymakers and early-stage investors.
Figure 1. SME-HGT architecture. Input features are projected per node type to a shared dimension, then refined through three HGT layers with residual connections. Only company node embeddings are passed to the classifier. A vertical flowchart showing five blocks connected by arrows from bottom to top: Heterogeneous Graph input, Type-Specific Linear Projection, three repeated HGT Layers with Residual connections and LayerNorm and Dropout, MLP Classifier for Company nodes, and Phase II Prediction output.
Model
AUPRC
AUROC
F1
XGBoost
0.629 ± .003
0.652 ± .003
0.590 ± .000
MLP
0.591 ± .025
0.634 ± .014
0.593 ± .001
R-GCN
0.598 ± .006
0.648 ± .006
0.593 ± .002
SME-HGT
0.606 ± .005
0.646 ± .003
0.591 ± .001
Table 2. Test Set Results (mean ± std over 10 seeds)
Precision@ K
Lift@ K
Model
@100
@500
@1K
@100
@500
@1K
XGBoost
.924
.711
.578
2.21
1.70
1.38
MLP
.845
.668
.550
2.02
1.60
1.32
R-GCN
.826
.673
.572
1.98
1.61
1.37
HGT
.834
.689
.576
2.00
1.65
1.38
Table 3. Ranking Metrics on Test Set (mean over 10 seeds)
Removed Feature
AUPRC
Δ vs. Full
None (full model)
0.603 ± .002
—
− log total Phase I
0.597 ± .002
− 0.006
− log Phase I count
0.577 ± .003
− 0.026
− agency diversity
0.604 ± .002
+ 0.001
− years active
0.607 ± .002
+ 0.005
− log avg Phase I size
0.601 ± .002
− 0.002
Table 4. Feature Ablation (HGT, AUPRC, mean ± std over 5 seeds). Larger Δ = more important feature.
Removed Edge Type
AUPRC
Δ vs. Full
None (full model)
0.603 ± .002
—
− operates_in
0.596 ± .003
− 0.007
− awarded_by
0.594 ± .005
− 0.009
− co_topic
0.615 ± .002
+ 0.012
− co_agency
0.605 ± .005
+ 0.002
− co_state
0.601 ± .006
− 0.002
Table 5. Edge-Type Ablation (HGT, AUPRC, mean ± std over 5 seeds).
Model
Edge Mode
AUPRC
AUROC
SME-HGT
Multi
0.606 ± .003
0.647 ± .003
Single
0.599 ± .004
0.645 ± .004
R-GCN
Multi
0.607 ± .007
0.653 ± .005
Single
0.594 ± .010
0.650 ± .007
MLP
Multi
0.557 ± .070
0.608 ± .051
Single
0.557 ± .070
0.608 ± .051
Table 6. Multi-Edge vs. Single-Edge (AUPRC, mean ± std over 5 seeds).
Model
Early
Current
Late
XGBoost
0.732 ± .001
0.628 ± .005
0.548 ± .005
MLP
0.660 ± .078
0.557 ± .070
0.433 ± .086
R-GCN
0.676 ± .059
0.607 ± .007
0.483 ± .080
SME-HGT
0.717 ± .004
0.606 ± .003
0.542 ± .005
Table 7. Temporal Robustness (AUPRC, mean ± std over 5 seeds). Test periods: Early (2018–2020), Current (2020–2022), Late (2021–2023).
Identifying the full landscape of small and medium-sized enterprises (SMEs) in specialized industry sectors is critical for supply-chain resilience, yet existing business databases suffer from substantial coverage gaps -- particularly for sub-tier suppliers and firms in emerging niche markets. We propose a \textbf{Web--Knowledge--Web (W→K→W)} pipeline that iteratively (1)~crawls domain-specific web sources to discover candidate supplier entities, (2)~extracts and consolidates structured knowledge into a heterogeneous knowledge graph using domain-adapted few-shot LLM prompting, and (3)~uses the knowledge graph's topology and coverage signals to guide subsequent crawling toward under-represented regions of the supplier space. To quantify discovery completeness, we introduce a coverage estimation framework inspired by ecological species-richness estimators (Chao1, ACE) adapted for web-entity populations. Experiments on the semiconductor equipment manufacturing sector (NAICS 333242) demonstrate that the W→K→W pipeline achieves the highest precision (0.165) and F1 (0.123) among all methods while using only 144 pages -- 32% fewer than the 213-page baseline budget -- building a knowledge graph of 664 entities and 542 relations with 100% relation type-consistency.
Yijiashun Qi, Yijiazhen Qi, Tanmay Wagh
University of Michigan Ann Arbor, USA · The University of Hong Kong Hong Kong, China · Santa Clara University Santa Clara, USA
Traditional decision tree models, which rely exclusively on numerical variables, often face challenges in handling high-dimensional data and are limited in their ability to incorporate textual information effectively. To address these limitations, we propose the integrated multivariate segmentation tree (IMST), a comprehensive framework designed to improve credit evaluation for small- and medium-sized enterprises (SMEs) by integrating financial data with textual sources. This method comprises three core stages: (1) transforming textual data into numerical matrices through matrix factorization, (2) selecting salient financial features using Lasso regression, and (3) constructing a multivariate segmentation tree based on either the Gini index or entropy, with weakest-link pruning applied to control model complexity. Experimental results based on a dataset of 1,428 Chinese SMEs demonstrated that IMST achieved an accuracy rate of 88.9%, surpassing both baseline decision trees (87.4%) and conventional models such as support vector machines and neural networks. Furthermore, the proposed model demonstrated superior interpretability and computational efficiency, featuring a more streamlined architecture and improved risk detection capabilities.
Lu Han, Xiuying Wang
School of Management Science and Engineering, Central University of Finance and Economics, Beijing 100081, China · School of Computer Science, University of Sydney, NSW 2006, Australia
Secondary private equity transactions require rapid synthesis of fragmented, unstructured financial disclosures, where critical metrics and their entity anchors are distributed across disjoint documents with limited lexical overlap. We present Aethel, a reproducible framework that combines bipartite Personalized PageRank graph retrieval with a coreference-aware Bipartite Coreference Teleportation layer and an orchestrated specialist-agent architecture. Aethel models corpora as entity-passage graphs and propagates relevance through explicit relational paths to support multi-hop financial diligence. We evaluate the retrieval layer on 200-question samples from the MuSiQue and 2WikiMultiHopQA validation sets, comparing sparse lexical, dense bi-encoder, vanilla graph, and coreference-aware graph retrieval. Aethel achieves HR@5 of 100.0% on 2WikiMultiHopQA and 88.5% on MuSiQue, improving coverage over vanilla graph retrieval while trading off top-rank precision. We also evaluate retrieval over a 4,123-chunk corpus of financial disclosures and find that graph retrieval outperforms dense retrieval on multi-hop recall but does not surpass a strong BM25 baseline at open-corpus scale. The results show that graph-based retrieval offers interpretable multi-hop evidence paths and degrades more gracefully than dense retrieval as corpus size grows, while also demonstrating that its advantage depends strongly on corpus scale and entity-index quality. Code and evaluation artifacts are released for reproducibility.