Yulong Ma

Yulong Ma

Xi'an University of Architecture and Technology (XAUAT)

I’m a Ph.D. candidate at the School of Management, Xi’an University of Architecture and Technology (XAUAT). My research focuses on mis/disinformation, AIGC, online public opinion, social intelligent governance, and NLP. I explore how AI-driven technologies can enhance the understanding and management of digital information ecosystems.

I obtained my bachelor’s degree from Anyang Normal University. I pursued a master’s degree at XAUAT. Now, I’m pursuing a doctoral degree at the Intelligent Management and Decision-Making Research (IMaDM) group in XAUAT, under the guidance of Prof. Xinsheng Zhang.

As of now, I have published 6 papers, participated in the publication of 9 papers, and authorized 1 software copyright. My English publications have accumulated [Getting Data from Google Scholar…] citations.

🔥 News

📝 Publications

Nature Communications (Q1, Top)
sym

Linguistic features of AI mis/disinformation and the detection limits of LLMs Yulong Ma, Xinsheng Zhang*, Jinge Ren, Runzhou Wang, Minghu Wang, Yang Chen

  • RQ1: Why do LLM-based linguistic feature fusion methods sometimes demonstrate limited effectiveness in detecting AI-generated mis/disinformation?
    • RQ1a: What are the similarities and differences between AI-generated real/fake content and human-authored content?
    • RQ1b: Which latent features could potentially enhance the detection accuracy of AI-generated mis/disinformation?
  • RQ2: Can LLMs reliably comprehend and distinguish AI-generated mis/disinformation under zero-shot prompt conditions?
    • RQ2a: Is there a difference in the intrinsic language comprehension patterns of LLMs when generating mis/disinformation?
    • RQ2b: What are the critical boundaries and limitations of LLMs’ capability to identify AI-generated mis/disinformation?
Applied Soft Computing (Q1, Top)
sym

Semantic-based topic model for public opinion analysis in sudden-onset disasters Yulong Ma, Xinsheng Zhang*, Runzhou Wang

  • Context representation problem of topic models is solved cost-effectively.
  • Manifold learning method is used to reconstruct semantics in low-dimensional space.
  • Weighted fusion methods between distribution vectors are explored.
  • Intrinsic processes of the NCTM are explained through visualization methods.
  • Results on three real datasets show that NCTM achieves SOTA performance.
Engineering Applications of Artificial Intelligence (Q1, Top)
sym

An ALBERT-based TextCNN-Hatt hybrid model enhanced with topic knowledge for sentiment analysis of sudden-onset disasters Xinsheng Zhang* (Supervisor), Yulong Ma

  • ALBERT is used as the backbone network to deal with the problem of conflicting emotions encountered in text generated during SODs.
  • The LDA model is used to integrate the topic knowledge of the external domain, which improves the model’s ability to perceive out-of-vocabulary words and implicit sentiments.
  • TextCNN and a hierarchical attention mechanism are used for feature reduction and attention, respectively.
  • Compared with previous SOTA models, the proposed model achieves more stable and better classification performance. The average accuracy and classification CQI obtained from 5-fold cross-validation were 88.79% and 87.80%, respectively.
  • The proposed model’s improved performance is analyzed using visual aids to establish the factors contributing to the improvements.

💬 Conferences

  • 2025 Nov. 01, The 10th (2025) Green Consumption and Green Marketing Symposium, Xi’an, Shaanxi, CN. Research on the Competition Pattern and Development Trend of China’s New Energy Vehicle Market: Based on Complex Network Analysis and Machine Learning.

  • 2025 Aug. 23, The 7th (2025) E-commerce and Digital Innovation Annual Conference, Huangshan, Anhui, CN. The Constraints of AI Misinformation Detection from the Perspectives of Fabrication Quality and Zero-Shot Prompting.

  • 2022 Oct. 15, INFORMS Workshop on Data Science, Indianapolis, Indiana, USA. An ALBERT-based TextCNN-Hatt hybrid model enhanced with topic features for sentiment analysis of sudden disasters.

🏆 Honors and Awards

  • 2026, 10, 2026 National Graduate Scholarship
  • 2025, 11, First-Class Graduate Academic Scholarship, Xi’an University of Architecture and Technology
  • 2025, 10, 2025 National Graduate Scholarship
  • 2025, 01, Outstanding Master’s Thesis Award, Xi’an University of Architecture and Technology
  • 2023, 12, First-Class Graduate Academic Scholarship, Xi’an University of Architecture and Technology
  • 2023, 09, Second-Class Graduate Academic Scholarship, Xi’an University of Architecture and Technology
  • 2023, 06, Outstanding Graduate, Xi’an University of Architecture and Technology
  • 2022, 12, First-Class Graduate Academic Scholarship, Xi’an University of Architecture and Technology
  • 2021, 12, First-Class Graduate Academic Scholarship, Xi’an University of Architecture and Technology
  • 2020, 11, First Prize, Anyang Social Science Planning Project
  • 2020, 06, Outstanding Graduate, Anyang Normal University

🥇 Competitions

  • 2025, 11, National Second Prize, 22nd China Postgraduate Mathematical Contest in Modeling
  • 2024, 12, National Second Prize, 21st China Postgraduate Mathematical Contest in Modeling (Top 1%)
  • 2023, 12, National Third Prize, 20th China Postgraduate Mathematical Contest in Modeling
  • 2021, 10, National Third Prize, 7th China International “Internet+” Innovation and Entrepreneurship Competition
  • 2021, 09, Provincial Gold Award and Rural Revitalization Special Award, 7th China International “Internet+” Innovation and Entrepreneurship Competition
  • 2021, 05, National Third Prize, “Charoen Pokphand Cup” 11th National Market Research and Analysis Competition (Graduate Group)
  • 2020, 11, National Third Prize, National English Competition for College Students (Graduate Group)
  • 2019, 12, Provincial First Prize, 11th National Undergraduate Mathematics Competition (Non-Mathematics Category)