Selected work

Representative Papers

RAGMesh with FaME-G2E: Long-Form Text-Driven 3D Face Generation and Editing

Aug 10, 2026

Existing methods struggle to accurately generate and edit fine-grained geometric details of 3D faces—such as eyebrow tension or cheek contraction—from long textual descriptions. To address this challenge, this work introduces FaME-G2E, a large-scale multimodal dataset, and proposes RAGMesh, a retrieval-augmented framework that integrates text-guided global and regional geometric priors in blendshape space. The framework innovatively combines a multi-scale retrieval fusion (MSRF) module with an adaptive RAG-guided supervision (AdaRAGS) mechanism to achieve precise semantic alignment and localized deformation control. Experimental results demonstrate that the proposed method significantly outperforms current state-of-the-art approaches in terms of local geometric accuracy, text controllability, regional editing precision, and inference efficiency.

0 citationsRead paper

How large is the error effect when summing or averaging nonlinear field normalization citation counts at the paper level?

Nov 04, 2025

This study investigates the measurement error introduced by summing or averaging citation counts after nonlinear field normalization and its implications for research impact assessment. Addressing the mathematical problem that nonlinear transformations violate metric isometry and distort aggregation, the authors systematically compare six linear and nonlinear normalization methods through empirical analysis on publication datasets from multiple universities, using raw citations and linearly normalized results as baselines. Results show that while aggregation after nonlinear normalization introduces bias, the overall error magnitude remains relatively small; critically, error amplification is strongly contingent on publication sample homogeneity—heterogeneous samples exhibit significantly larger errors. This work provides the first quantitative characterization of the interaction effect between normalization choice and aggregation operation, establishing a foundational methodological framework for selecting, interpreting, and justifying normalization strategies in scientometric evaluation.

0 citationsRead paper

Mathematical proof concerning the additivity problem of nonlinear normalized citation counts

Aug 20, 2025

The additivity of nonlinearly normalized citation counts in scientometrics—critical for data integration and interpretation—remains theoretically unresolved. Method: We conduct a rigorous functional analysis over the real numbers, employing proof by contradiction and theorem derivation to examine whether any continuous or monotonic nonlinear normalization preserves isometry. Contribution/Results: We establish, for the first time, that all such nonlinear normalizations necessarily violate isometry, thereby rendering citation counts non-additive. This result holds universally—not only for mainstream normalization techniques (e.g., field-normalized or percentile-based methods) but also for broader nonlinear data transformations across multidisciplinary contexts. Our findings expose a fundamental limitation of nonlinear processing in scientific evaluation, providing a key theoretical caution for citation analytics methodology. They further advocate a methodological shift toward additive, comparable frameworks—specifically, linear or structurally constrained modeling approaches—that preserve metric integrity and enable robust cross-study aggregation.

0 citationsRead paper

Recent publications

Latest Papers

RAGMesh with FaME-G2E: Long-Form Text-Driven 3D Face Generation and Editing

Aug 10, 2026

Existing methods struggle to accurately generate and edit fine-grained geometric details of 3D faces—such as eyebrow tension or cheek contraction—from long textual descriptions. To address this challenge, this work introduces FaME-G2E, a large-scale multimodal dataset, and proposes RAGMesh, a retrieval-augmented framework that integrates text-guided global and regional geometric priors in blendshape space. The framework innovatively combines a multi-scale retrieval fusion (MSRF) module with an adaptive RAG-guided supervision (AdaRAGS) mechanism to achieve precise semantic alignment and localized deformation control. Experimental results demonstrate that the proposed method significantly outperforms current state-of-the-art approaches in terms of local geometric accuracy, text controllability, regional editing precision, and inference efficiency.

0 citationsRead paper

How large is the error effect when summing or averaging nonlinear field normalization citation counts at the paper level?

Nov 04, 2025

This study investigates the measurement error introduced by summing or averaging citation counts after nonlinear field normalization and its implications for research impact assessment. Addressing the mathematical problem that nonlinear transformations violate metric isometry and distort aggregation, the authors systematically compare six linear and nonlinear normalization methods through empirical analysis on publication datasets from multiple universities, using raw citations and linearly normalized results as baselines. Results show that while aggregation after nonlinear normalization introduces bias, the overall error magnitude remains relatively small; critically, error amplification is strongly contingent on publication sample homogeneity—heterogeneous samples exhibit significantly larger errors. This work provides the first quantitative characterization of the interaction effect between normalization choice and aggregation operation, establishing a foundational methodological framework for selecting, interpreting, and justifying normalization strategies in scientometric evaluation.

0 citationsRead paper

Mathematical proof concerning the additivity problem of nonlinear normalized citation counts

Aug 20, 2025

The additivity of nonlinearly normalized citation counts in scientometrics—critical for data integration and interpretation—remains theoretically unresolved. Method: We conduct a rigorous functional analysis over the real numbers, employing proof by contradiction and theorem derivation to examine whether any continuous or monotonic nonlinear normalization preserves isometry. Contribution/Results: We establish, for the first time, that all such nonlinear normalizations necessarily violate isometry, thereby rendering citation counts non-additive. This result holds universally—not only for mainstream normalization techniques (e.g., field-normalized or percentile-based methods) but also for broader nonlinear data transformations across multidisciplinary contexts. Our findings expose a fundamental limitation of nonlinear processing in scientific evaluation, providing a key theoretical caution for citation analytics methodology. They further advocate a methodological shift toward additive, comparable frameworks—specifically, linear or structurally constrained modeling approaches—that preserve metric integrity and enable robust cross-study aggregation.

0 citationsRead paper