Zhaoqing Wu

Author directory

2026

In recent years, there has been a surge of interest in Cultural NLP, with substantial efforts to create globally inclusive NLP systems. The rapid growth of literature in this field makes it difficult to track trends in methods and data resources. To address this, we survey over 375 papers to answer three complementary questions: (1) What Cultural Capabilities (CCs) are being targeted in NLP systems? (2) How are cultural data resources being created? and (3) What methods are being used to improve the CCs of those systems? We discuss trends observed across the three questions, and identify relevant research gaps. To facilitate further research in this field, we release our full list of surveyed papers, in the form of an interactive web interface, CultureMine, which includes a feature to allow researchers to add their work; we hope this facilitates future research and proves to be a valuable resource for the Cultural NLP community.
Large language models (LLMs) demonstrate strong instruction-following and reasoning abilities, yet struggle with the socially nuanced understanding humans handle naturally. Humans rely on social sense to interpret intentions, emotions, and relationships in context. We investigate whether and to what extent LLMs exhibit similar abilities when reasoning about social interactions. We propose a structured approach that decomposes social interactions into three aspects and prompts LLMs to generate aspect-based summaries capturing complementary signals. We evaluate these summaries and resulting reasoning within a unified framework across three dimensions, analyzing both LLM–human alignment and the quality of LLM-generated social explanations. Our results show that LLMs achieve moderate agreement with humans and with one another as evaluators and capture some socially nuanced information across aspects, but struggle with interaction-level dynamics shaped by situational norms. Across four downstream CSS tasks, aspect summaries improve aggregate prediction performance across three model families, though effects vary by task. Complementary rationale analysis shows that summaries substantially alter expressed reasoning. Finally, we introduce a benchmark dataset with gold-standard annotations to support systematic study of social understanding and situated reasoning in LLMs.

2025

Large language models (LLMs) struggle in social science domains, where critical thinking and human-level inference are crucial. In this work, we propose a multi-agent social reasoning framework that leverages the generative and reasoning capabilities of LLMs to generate and evaluate reasons from multiple perspectives grounded in social science theories, and construct a factor graph for inference. Experimental results on understanding power dynamics in conversations show that our method outperforms standard prompting baselines, demonstrating its potential for tackling hard Computational Social Science (CSS) tasks.