arXiv subjects

Naipeng Chao

Publications and source records attributed to Naipeng Chao.

3 recordsLinked to original sources

arXiv2026

PersonaArena: Dynamic Simulation for Evaluating and Enhancing Persona-Level Role-Playing in Large Language Models

Large language models (LLMs) increasingly serve as interactive social agents, yet their ability to maintain coherent and authentic persona-level role-playing remains limited, particularly in realistic social scenarios. Existing research predominantly focuses on character-level settings and relies on static evaluation formats, failing to capture the complexity of everyday social interactions. In this work, we present PersonaArena, a dynamic simulation framework for evaluating and improving persona-level role-playing in LLMs. PersonaArena leverages a large, filtered corpus of user-generated social content to construct a nuanced persona bank, and elicits multi-turn, context-rich interactions within simulated social environments. Our framework features a multi-agent debating judge for holistic and unbiased assessment. Through extensive experiments, we demonstrate that PersonaArena enables rigorous evaluation and enhancement of LLMs' role-playing capabilities, advancing the development of more authentic and socially adept AI agents.

cs.AI↗

arXiv2026

Toward Consistent World Models with Multi-Token Prediction and Latent Semantic Enhancement

Whether Large Language Models (LLMs) develop coherent internal world models remains a core debate. While conventional Next-Token Prediction (NTP) focuses on one-step-ahead supervision, Multi-Token Prediction (MTP) has shown promise in learning more structured representations. In this work, we provide a theoretical perspective analyzing the gradient inductive bias of MTP, supported by empirical evidence, showing that MTP promotes the convergence toward internal belief states by inducing representational contractivity via gradient coupling. However, we reveal that standard MTP often suffers from structural hallucinations, where discrete token supervision encourages illegal shortcuts in latent space that violate environmental constraints. To address this, we propose a novel method Latent Semantic Enhancement MTP (LSE-MTP), which anchors predictions to ground-truth hidden state trajectories. Experiments on synthetic graphs and real-world Manhattan Taxi Ride show that LSE-MTP effectively bridges the gap between discrete tokens and continuous state representations, enhancing representation alignment, reducing structural hallucinations, and improving robustness to perturbations.

cs.LG↗

arXiv2017

Leveraging the Flow of Collective Attention for Computational Communication Research

Human attention becomes an increasingly important resource for our understanding or collective human behaviors in the age of information explosion. To better understand the flow of collective attention, we construct the attention flow network using anonymous smartphone data of 100,000 users in a major city of China. In the constructed network, nodes are websites visited by users, and links denote the switch of users between two websites. We quantify the flow of collective attention by computing the flow network statistics, such as flow impact, flow dissipation, and flow distance. The findings reveal a strong concentration and fragmentation of collective attention for smartphone users, while the duplication of attention cross websites proves to be unfounded in mobile using. We further confirmed the law of dissipation and the allowmetric scaling of flow impact. Surprisingly, there is a centralized flow structure, suggesting that the website with large traffic can easily control the circulated collective attention. Additionally, we find that flow network analysis can effectively explain the page views and sale volume of products. Finally, we discuss the benefits and limitations of using the flow network analysis for computational communication research.

cs.CY↗