Xiaoyu Li

Dr. Xiaoyu Li

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 September 2024 in Information Processing & Management
Verified
Journal Article
Closed access
XLXiaoyu Li
Xiaoyu Li
SGShaoyang Guo
Shaoyang Guo
JWJin Wu
Jin Wu
CZChanjin Zheng
Chanjin Zheng

1 October 2024 in British Journal of Mathematical and Statistical Psychology
Verified
Journal Article
Closed access
XLXiaoyu Li
Xiaoyu Li
SDShenghong Dong
Shenghong Dong
SGShaoyang Guo
Shaoyang Guo
CZChanjin Zheng
Chanjin Zheng

7 July 2026 in Behavioral Sciences
Verified
Journal Article
Open access
PDF file
XLXiaoyu Li
Xiaoyu Li
JWJin Wu
Jin Wu
YRYupei Ren
Yupei Ren
SGShaoyang Guo
Shaoyang Guo
ZLZhongquan Li
Zhongquan Li
CZChanjin Zheng
Chanjin Zheng
In the era of intelligent assessment, psychometric tests are becoming increasingly important for personnel selection, career development, and mental health assessment. Forced-choice tests are common in personality assessments because they require participants to select from closely related options, lowering the risk of response distortion. However, traditional latent trait models for forced-choice tests suffer from severe computational bottlenecks in high-dimensional settings. Furthermore, existing deep learning-based cognitive diagnosis models are primarily designed for independent items in educational scenarios (predicated on absolute scoring), making them structurally maladapted to the ipsative data (relative preference comparisons) generated by forced-choice tests. To address these challenges, this study presents a deep learning-based Forced-Choice Neural Latent Trait (FCNLT) Model that overcomes the limitations of traditional models and is applicable to the three most common item block types found in forced-choice tests. To account for the unidimensionality of items, participants’ latent trait levels and item characteristics are represented as interpretable latent embeddings. FCNLT mines these features through nonlinear mapping and introduces a weighted BPR-based ranking loss to natively align with the relative-scoring nature of forced-choice data. Additionally, the monotonicity assumption is utilized to improve the interpretability of the trait estimates. The FCNLT’s effectiveness is validated by experiments on real-world and simulated datasets that show its accuracy, interpretability, and robustness.

1 September 2026 in Journal of Intelligence
Verified
Journal Article
Open access
PDF file
SGShaoyang Guo
Shaoyang Guo
YXYinmei Xu
Yinmei Xu
XYXiangdong Yang
Xiangdong Yang
YCYanlei Chen
Yanlei Chen
XLXiaoyu Li
Xiaoyu Li
XTXuefei Tian
Xuefei Tian
Collaborative problem-solving (CPS) is an essential 21st-century competency, yet its valid and ecologically sound assessment remains challenging due to inconsistent frameworks, poorly aligned task design, and over-reliance on behavior frequency counting in scoring. This study developed and provided empirical support for a theory-driven game-based CPS assessment embedded in a real-time strategy game, using a customized Collaborative Siege Task with asymmetric roles and ill-structured problems, to elicit more realistic collaborative interactions. Based on a validated three-facet CPS framework from Chinese cultural context, the task systematically triggers social and cognitive processes of collaboration across five progressive stages. By sampling from the eastern provinces of China, a pilot study with 36 participants confirmed task feasibility and indicator effectiveness, while a formal study with 298 valid participants examined psychometric properties using item response theory and confirmatory factor analysis. Findings supported the hierarchical higher-order factor structure, satisfactory item difficulty, discrimination, reliability, and criterion-related validity against peer assessment, reasoning ability, and personality. The assessment showed no sex, major, or role bias. This study provides a reliable gamified assessment tool for CPS measurement with demonstrated potential for scalability pending further development of automated scoring procedures, offering implications for large-scale evaluation and competency-oriented instruction.

30 March 2026 in Sustainability
Verified
Journal Article
Open access
PDF file
YRYupei Ren
Yupei Ren
NZNing Zhang
Ning Zhang
XLXiaoyu Li
Xiaoyu Li
YZYadong Zhang
Yadong Zhang
YCYuqing Chen
Yuqing Chen
MLMan Lan
Man Lan
As critical elements in argumentative writing, argument components and strategies significantly influence argument quality. However, the existing research lacks an in-depth exploration of how students construct and utilize these elements in argumentative writing. This study first evaluates the performance of leading large language models (LLMs) in identifying argument components and strategies using three approaches: single-task learning (STL), chain-of-thought (CoT), and multi-task learning (MTL). With the aid of learning analytics methods (Epistemic Network Analysis (ENA) and two-mode network), the study further reveals the intrinsic mechanisms linking argument components, strategies, and writing quality. Specifically, the research trains and evaluates LLMs on 226 argumentative essays, encompassing 4726 components and 4837 strategies. Compared to basic STL, the CoT and MTL methods significantly improve LLMs’ performance in both tasks. Moreover, learning analytics indicate that high-quality essays possess rich and complex logical relations, presenting multidimensional and multi-layered reasoning structures, whereas low-quality essays predominantly rely on simple and repetitive connections, lacking deeper logical support. These findings have significant implications for the automated analysis of argumentative writing and the sustainable development of education, not only providing valuable insights for educators in argumentation instruction but also contributing to the systematic enhancement of students’ argumentative abilities and critical thinking.

20 July 2025 in arXiv
Verified
Preprint
Open access
PDF file
XLXiaoyu Li
Xiaoyu Li
JWJin Wu
Jin Wu
SGShaoyang Guo
Shaoyang Guo
HSHaoran Shi
Haoran Shi
CZChanjin Zheng
Chanjin Zheng
In the smart era, psychometric tests are becoming increasingly important for personnel selection, career development, and mental health assessment. Forced-choice tests are common in personality assessments because they require participants to select from closely related options, lowering the risk of response distortion. This study presents a deep learning-based Forced-Choice Neural Cognitive Diagnostic Model (FCNCD) that overcomes the limitations of traditional models and is applicable to the three most common item block types found in forced-choice tests. To account for the unidimensionality of items in forced-choice tests, we create interpretable participant and item parameters. We model the interactions between participant and item features using multilayer neural networks after mining them using nonlinear mapping. In addition, we use the monotonicity assumption to improve the interpretability of the diagnostic results. The FCNCD's effectiveness is validated by experiments on real-world and simulated datasets that show its accuracy, interpretability, and robustness.