Selected work

Representative Papers

FactSim: Fact-Checking for Opinion Summarization

Feb 09, 2026

This work addresses the challenge that existing automatic evaluation metrics struggle to accurately assess the factual consistency of opinion summaries generated by large language models. To this end, the authors propose FactSim, an end-to-end fully automated evaluation method that extracts factual claims from both the generated summary and the original user reviews, and introduces a robust fact similarity scoring mechanism designed to handle negations, paraphrases, and elaborations. This approach effectively measures both factual consistency and coverage. Experimental results demonstrate that FactSim achieves significantly higher correlation with human judgments than current state-of-the-art automatic metrics, offering a more reliable proxy for human evaluation in assessing factual faithfulness.

0 citationsRead paper

Verification of Lightning Network Channel Balances with Trusted Execution Environments (TEE)

Dec 12, 2025

This work addresses the fundamental tension between channel liquidity state privacy and third-party verifiability in the Lightning Network (LN). We propose the first dual-assurance verification framework integrating Trusted Execution Environment (TEE) remote attestation with zero-knowledge Transport Layer Security (zkTLS). Methodologically, we design a synergistic mechanism of Hot Proofs—real-time enclave-resident balance attestations leveraging SGX or SEV—and Cold Proofs—on-chain settlement records—precisely delineating security boundaries and trade-offs. We further implement a Balanced Reporting API and zkTLS-based transport-layer verification to jointly ensure hardware-enforced integrity and end-to-end communication privacy. Our key contribution is the first deep integration of TEEs and zkTLS for off-chain liquidity auditing, eliminating risks of node tampering and third-party API leakage. Empirical evaluation demonstrates bounded verification latency, enabling high-confidence, non-intrusive financial capacity verification. The framework delivers a deployable, trustworthy verification infrastructure for LN auditors, service providers, and node operators.

0 citationsRead paper

Revisiting Graph Projections for Effective Complementary Product Recommendation

Jun 10, 2025

Complementary product recommendation suffers from severe data sparsity and noise in user–item interaction graphs. To address the inadequate modeling of complementary relationships in bipartite graphs, this work systematically redefines the bipartite graph projection paradigm. We propose a directed weighted projection method to explicitly capture the directional nature of complementarity; design an edge-weight recalibration mechanism to suppress noise-induced distortions; and introduce structured neighborhood aggregation coupled with a parameter-free similarity propagation strategy to enhance relational generalization under sparsity. Evaluated on multiple benchmark datasets, our approach achieves average improvements of 43% and 38% in recommendation accuracy over state-of-the-art sequential and graph neural recommender models, respectively. It significantly advances complementary item identification performance under sparse and noisy conditions.

0 citationsRead paper

ROSA: Addressing text understanding challenges in photographs via ROtated SAmpling

Jun 04, 2025

Existing vision-language models for visual question answering (VQA) suffer significant performance degradation on images captured by visually impaired users—characterized by skewed text orientation and off-center composition—due to the predominant reliance of mainstream benchmarks on upright, well-framed text images, which inadequately reflect real-world accessibility scenarios. Method: This paper introduces the first cognition-aware framework grounded in empirical analysis of visually impaired users’ image-capturing behaviors. We propose a lightweight rotation-aware sampling and decoding strategy that robustly models text orientation without architectural modification. Specifically, it performs multi-angle image sampling followed by fused decoding to enhance comprehension of arbitrarily oriented text. Contribution/Results: On text-dense images, our method achieves an absolute accuracy gain of 11.7 percentage points over standard greedy decoding. It substantially improves VQA’s practical utility and generalization capability in accessibility-critical settings, establishing a new direction for inclusive multimodal reasoning.

0 citationsRead paper

Recent publications

Latest Papers

FactSim: Fact-Checking for Opinion Summarization

Feb 09, 2026

This work addresses the challenge that existing automatic evaluation metrics struggle to accurately assess the factual consistency of opinion summaries generated by large language models. To this end, the authors propose FactSim, an end-to-end fully automated evaluation method that extracts factual claims from both the generated summary and the original user reviews, and introduces a robust fact similarity scoring mechanism designed to handle negations, paraphrases, and elaborations. This approach effectively measures both factual consistency and coverage. Experimental results demonstrate that FactSim achieves significantly higher correlation with human judgments than current state-of-the-art automatic metrics, offering a more reliable proxy for human evaluation in assessing factual faithfulness.

0 citationsRead paper

Verification of Lightning Network Channel Balances with Trusted Execution Environments (TEE)

Dec 12, 2025

This work addresses the fundamental tension between channel liquidity state privacy and third-party verifiability in the Lightning Network (LN). We propose the first dual-assurance verification framework integrating Trusted Execution Environment (TEE) remote attestation with zero-knowledge Transport Layer Security (zkTLS). Methodologically, we design a synergistic mechanism of Hot Proofs—real-time enclave-resident balance attestations leveraging SGX or SEV—and Cold Proofs—on-chain settlement records—precisely delineating security boundaries and trade-offs. We further implement a Balanced Reporting API and zkTLS-based transport-layer verification to jointly ensure hardware-enforced integrity and end-to-end communication privacy. Our key contribution is the first deep integration of TEEs and zkTLS for off-chain liquidity auditing, eliminating risks of node tampering and third-party API leakage. Empirical evaluation demonstrates bounded verification latency, enabling high-confidence, non-intrusive financial capacity verification. The framework delivers a deployable, trustworthy verification infrastructure for LN auditors, service providers, and node operators.

0 citationsRead paper

Revisiting Graph Projections for Effective Complementary Product Recommendation

Jun 10, 2025

Complementary product recommendation suffers from severe data sparsity and noise in user–item interaction graphs. To address the inadequate modeling of complementary relationships in bipartite graphs, this work systematically redefines the bipartite graph projection paradigm. We propose a directed weighted projection method to explicitly capture the directional nature of complementarity; design an edge-weight recalibration mechanism to suppress noise-induced distortions; and introduce structured neighborhood aggregation coupled with a parameter-free similarity propagation strategy to enhance relational generalization under sparsity. Evaluated on multiple benchmark datasets, our approach achieves average improvements of 43% and 38% in recommendation accuracy over state-of-the-art sequential and graph neural recommender models, respectively. It significantly advances complementary item identification performance under sparse and noisy conditions.

0 citationsRead paper

ROSA: Addressing text understanding challenges in photographs via ROtated SAmpling

Jun 04, 2025

Existing vision-language models for visual question answering (VQA) suffer significant performance degradation on images captured by visually impaired users—characterized by skewed text orientation and off-center composition—due to the predominant reliance of mainstream benchmarks on upright, well-framed text images, which inadequately reflect real-world accessibility scenarios. Method: This paper introduces the first cognition-aware framework grounded in empirical analysis of visually impaired users’ image-capturing behaviors. We propose a lightweight rotation-aware sampling and decoding strategy that robustly models text orientation without architectural modification. Specifically, it performs multi-angle image sampling followed by fused decoding to enhance comprehension of arbitrarily oriented text. Contribution/Results: On text-dense images, our method achieves an absolute accuracy gain of 11.7 percentage points over standard greedy decoding. It substantially improves VQA’s practical utility and generalization capability in accessibility-critical settings, establishing a new direction for inclusive multimodal reasoning.

0 citationsRead paper