Papers by Qinghua Zhao
KESA: A Knowledge Enhanced Approach To Sentiment Analysis (2022.aacl-main)
Copied to clipboard
| Challenge: | Recent work on injecting sentiment knowledge into pre-trained language models, but it is difficult to integrate external knowledge into PLMs. |
| Approach: | They propose two sentiment-aware auxiliary tasks to integrate sentiment knowledge into the objective of the downstream task. |
| Outcome: | The proposed tasks outperform baselines and complement existing sentiment-enhanced models. |
SynapticRAG: Enhancing Temporal Memory Retrieval in Large Language Models through Synaptic Mechanisms (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing retrieval methods in Large Language Models show degradation in accuracy when handling temporally distributed conversations. |
| Approach: | They propose a method that combines temporal triggers and synaptic-like stimulus propagation to identify relevant dialogue histories. |
| Outcome: | The proposed approach improves on four datasets of English, Chinese and Japanese compared to state-of-the-art retrieval methods by 14.66% points. |
Dynamic Personality in LLM Agents: A Framework for Evolutionary Modeling and Behavioral Analysis in the Prisoner’s Dilemma (2025.findings-acl)
Copied to clipboard
| Challenge: | Current models rely on static personality traits but lack natural selection processes and direct psychological metrics, failing to accurately capture authentic dynamic personality variations. |
| Approach: | They propose a framework that uses game payoffs as environmental feedback to drive adaptive personality evolution and analyze correlations between personality metrics and behavior. |
| Outcome: | The proposed framework reveals new behavioral patterns of agents and evaluates personality-behavior relationships, advancing agent-based social simulations and human-AI symbiosis research. |
A Layer-wise Analysis of Supervised Fine-Tuning (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning ignore depth-dependent heterogeneity of instruction-following . a critical gap remains in understanding where these changes occur across the model's depth and which layers are essential for instruction- following. |
| Approach: | They propose a method which selectively updates critical intermediate layers . they show that effective alignment is architecturally localized rather than distributed . |
| Outcome: | The proposed method outperforms standard LoRA up to 10.2% on GSM8K with reduced parameter overhead. |
Entity Concept-enhanced Few-shot Relation Extraction (2021.acl-short)
Copied to clipboard
| Challenge: | Existing FSRE methods fail to classify relations based on information of sentences and entity pairs due to limited samples and lack of knowledge. |
| Approach: | They propose a concept-sentence attention module to select the most appropriate concept from multiple concepts of each entity by calculating the semantic similarity between sentences and concepts. |
| Outcome: | The proposed scheme outperforms existing methods on a few-shot relation extraction dataset. |
How Chain-of-Thought Works? Tracing Information Flow from Decoding, Projection, and Activation (2026.findings-acl)
Copied to clipboard
| Challenge: | Chain-of-Thought (CoT) prompting significantly enhances model reasoning, yet its internal mechanisms remain poorly understood. |
| Approach: | They reversely traced information flow across decoding, projection, and activation phases and found that CoT may serve as a decoding space pruner . |
| Outcome: | The proposed framework can be used to design more efficient and robust prompts. |