Papers by Xiaoyi Wang

21 papers
Towards Storage-Efficient Visual Document Retrieval: An Empirical Study on Reducing Patch-Level Embeddings (2025.findings-acl)

Copied to clipboard

Challenge: Visualized Document Retrieval (VDR) uses large vision-language models to encode document pages into embeddings.
Approach: They evaluate methods to reduce patch embeddings per page while minimizing performance degradation.
Outcome: The proposed method maintains 98.2% of retrieval performance with only 11.8% of original memory usage and preserves 94.6% effectiveness at 2% memory footprint.
InfiniteWeb: Scalable Web Environment Synthesis for GUI Agent Training (2026.acl-long)

Copied to clipboard

Challenge: Existing GUI agent benchmarks are manually constructed and lack scale and diversity as training environments.
Approach: They propose a GUI agent training system that automatically generates web environments at scale.
Outcome: The proposed system outperforms commercial GUI agents at realistic website construction and improves on OSWorld and Online-Mind2Web.
Exploring Knowledge Filtering for Retrieval-Augmented Discriminative Tasks (2025.findings-acl)

Copied to clipboard

Challenge: Recent studies have focused on generative tasks, while its potential in discriminative tasks remains largely unexplored.
Approach: They propose a framework that incorporates knowledge filtering and prediction fusion mechanisms to improve model performance.
Outcome: The proposed framework improves model performance on discriminative tasks by filtering out harmful knowledge and integrating it into the input context.
From Off-Policy to On-Policy: Enhancing GUI Agents via Bi-level Expert-to-Policy Assimilation (2026.acl-long)

Copied to clipboard

Challenge: Vision-language models are increasingly deployed as computer-use agents that operate desktops and browsers.
Approach: They propose a method that turns static expert traces into policy-aligned guidance . they propose RLVR with a per-task, dynamically updated cache to decompose planning and execution .
Outcome: The proposed model improves UITARS1.5-7B success from 22.87% to 32.13% on OSWorld-Verified and raises a held-out split from 5.74% to 10.30% on MMBench-GUI and Online-Mind2Web.
Improving Preference Alignment of LLM with Inference-Free Self-Refinement (2025.findings-emnlp)

Copied to clipboard

Challenge: Large language models (LLMs) develop in-context learning capability through pretraining and instruction tuning.
Approach: Large language models (LLMs) develop in-context learning capability through pretraining and instruction tuning.
Outcome: Experiments show that incorporating IFSR into preference alignment yields performance improvement over 10%.
SongComposer: A Large Language Model for Lyric and Melody Generation in Song Composition (2025.acl-long)

Copied to clipboard

Challenge: Creating lyrics and melodies in symbolic format requires expert knowledge of melody and an advanced understanding of lyrics.
Approach: They introduce SongComposer, a music-specialized large language model that can create symbolic lyrics and melodies following instructions.
Outcome: The proposed model outperforms existing models in symbolic song composition tasks.
Exploring Hybrid Sampling Inference for Aspect-based Sentiment Analysis (2025.findings-naacl)

Copied to clipboard

Challenge: Existing methods for inference require multiple sampling with preset size . however, it is a high-cost method that requires multiple sampling .
Approach: They propose a method that combines multiple and single sampling to greatly reduce the cost of multiple sampling without sacrificing performance.
Outcome: The proposed method greatly reduces the cost of multiple sampling without sacrificing performance.
Towards Robust Few-Shot Relation Classification: Incorporating Relation Description with Agreement (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing approaches to recognize relational relationships with a few support samples are limited for unlimited queries.
Approach: They propose a simple but effective framework that uses relation descriptions as external knowledge to enhance the model’s comprehension of the relation semantics.
Outcome: The proposed framework outperforms strong baselines while being robust against various NOTA rates.
ProductResearch: Training E-Commerce Deep Research Agents via Multi-Agent Synthetic Trajectory Distillation (2026.acl-industry)

Copied to clipboard

Challenge: Existing LLM-based agents lack the interaction depth and contextual breadth required for complex product research.
Approach: They propose a multi-agent framework that synthesizes high-fidelity tool-use trajectories for training robust e-commerce shopping agents.
Outcome: The proposed framework synthesizes high-fidelity, long-horizon tool-use trajectories for training robust e-commerce shopping agents.
Sentimental Image Generation for Aspect-based Sentiment Analysis (2025.findings-acl)

Copied to clipboard

Challenge: Recent work on textual Aspect-Based Sentiment Analysis (ABSA) has demonstrated promising performance, but limited semantics derived from raw data.
Approach: They propose a method that provides visual semantics to reinforce textual ABSA by adding additional augmentations to the input data.
Outcome: The proposed method can provide visual semantics to reinforce the textual extraction.
Exploring Graph Pre-training for Aspect-based Sentiment Analysis (2023.findings-emnlp)

Copied to clipboard

Challenge: Existing studies tend to extract the sentiment elements in a generative manner to avoid complex modeling of sentiment elements.
Approach: They propose a generative model with an Element-level Graph Pre-training paradigm and a Task Decomposition Pre- training paradigm to make it generalizable and robust against irregular sentiment quadruples.
Outcome: The proposed model is generalizable and robust against irregular sentiment quadruples.
Simple but Effective Compound Geometric Operations for Temporal Knowledge Graph Completion (2024.acl-long)

Copied to clipboard

Challenge: Current methods embed factual knowledge into continuous vector space and apply geometric operations to learn potential patterns in temporal knowledge graphs.
Approach: They propose a temporal knowledge graph completion method that uses two geometric operations to learn missing facts in temporal graphs.
Outcome: The proposed method significantly outperforms existing temporal knowledge graph embedding models.
InternLM-XComposer2.5-Reward: A Simple Yet Effective Multi-Modal Reward Model (2025.findings-acl)

Copied to clipboard

Challenge: Despite the promising performance of Large Vision Language Models, they sometimes generate incorrect outputs.
Approach: They propose a multi-modal reward model that aligns LVLMs with human preferences.
Outcome: The proposed model achieves excellent results on the latest multi-modal reward model benchmark and shows competitive performance on text-only reward model.
Revisiting Classical Chinese Event Extraction with Ancient Literature Information (2025.acl-long)

Copied to clipboard

Challenge: Existing studies on classical Chinese event extraction focus on grafting the complex modeling from English or modern Chinese works, neglecting the unique characteristic of this language.
Approach: They propose a Literary Vision-Language Model (VLM) for classical Chinese event extraction . they integrate annotations, historical background and character glyphs to capture the inner- and outer-context information from the sequence.
Outcome: The proposed model can capture the inner- and outer-context information at nearly zero cost.
Opinion Tree Parsing for Aspect-based Sentiment Analysis (2023.findings-acl)

Copied to clipboard

Challenge: Existing generative models for aspect-based sentiment analysis lack structure well-formedness guarantees and built-in elements alignments.
Approach: They propose an opinion tree parsing model which parses all sentiment elements from an opinion-tree.
Outcome: The proposed model is much faster than previous models and can explore correlations among sentiment elements.
Employing Glyphic Information for Chinese Event Extraction with Vision-Language Model (2024.findings-emnlp)

Copied to clipboard

Challenge: Recent studies on event extraction have incorporated a variety of features, including textual elements and annotations.
Approach: They propose a glyphic multi-modal Chinese event extraction model with hieroglyphic images to capture morphological structure from the sequence.
Outcome: The proposed model can extract events from a Chinese and KBP Eval datasets at low cost.
Improve Decoding Factuality by Token-wise Cross Layer Entropy of Large Language Models (2025.findings-naacl)

Copied to clipboard

Challenge: Large language models (LLMs) often struggle with the issue of generating inaccurate or fabricated content even when they possess correct knowledge.
Approach: They propose a decoding method that mitigates hallucinations without extra training . they propose entropy eNhanced decoding that leverages inner probability changes .
Outcome: The proposed method improves the truthfulness and informativeness of generation while maintaining robust QA accuracy.
Feel the Difference? A Comparative Analysis of Emotional Arcs in Real and LLM-Generated CBT Sessions (2025.findings-emnlp)

Copied to clipboard

Challenge: Synthetic therapy dialogues generated by large language models (LLMs) lack the nuanced emotional dynamics of real therapy.
Approach: They introduce a dataset of authentic cognitive behavioral therapy dialogues and analyze emotional arcs between real and LLM-generated CBT sessions.
Outcome: The proposed dataset is a comparative analysis of emotional arcs between real and LLM-generated CBT sessions.
CalligraphicOCR for Chinese Calligraphy Recognition (2025.emnlp-main)

Copied to clipboard

Challenge: Increasing efforts to digitize calligraphy have rely on isolated character recognition, requiring expensive manual splitting into single characters.
Approach: They propose a calligraphicOCR model with calligraphy image augmentation and action-based corrector targeting the root of the problem.
Outcome: The proposed model outperforms baseline models due to visual variations and domain shifts in semantics and is more accurate than previous models.
ECoK: Emotional Commonsense Knowledge Graph for Mining Emotional Gold (2024.findings-acl)

Copied to clipboard

Challenge: Existing knowledge graphs focus on the representation and reasoning of general factual knowledge, while there are significant deficiencies in the understanding and reasoning for emotional knowledge.
Approach: They propose a commonsense knowledge graph that can be used to represent emotional knowledge by combining theories from psychology, cognitive science, and linguistics.
Outcome: The proposed model surpasses GPT-4-Turbo in the emotion-related tasks.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations