Papers by Yifan Song

33 papers
Hierarchical Memory Organization for Wikipedia Generation (2025.acl-long)

Copied to clipboard

Challenge: Existing methods for generating Wikipedia articles do not utilize memory directly for outline generation.
Approach: They propose a method to generate Wikipedia articles autonomously by leveraging a hierarchical memory architecture.
Outcome: The proposed framework outperforms baseline methods in producing informative and reliable articles.
CodeRAG: Finding Relevant and Necessary Knowledge for Retrieval-Augmented Repository-Level Code Completion (2025.emnlp-main)

Copied to clipboard

Challenge: Recent advances in code large language models have produced repository-level code completion methods that automatically predict the unfinished code based on the broader information from the repository.
Approach: They propose a framework to identify relevant knowledge for retrieval-augmented repository-level code completion.
Outcome: The proposed framework significantly outperforms state-of-the-art methods on ReccEval and CCEval.
Complex Numerical Reasoning with Numerical Semantic Pre-training Framework (2025.emnlp-main)

Copied to clipboard

Challenge: Numerical knowledge graphs (NKGs) are not limited to discrete entity-relation knowledge.
Approach: They propose to combine numerical values and entities to solve multi-hop complex reasoning over incomplete knowledge graphs.
Outcome: The proposed approach handles up to 102 types of complex numerical reasoning queries on three public datasets.
AgentBank: Towards Generalized LLM Agents via Fine-Tuning on 50000+ Interaction Trajectories (2024.findings-emnlp)

Copied to clipboard

Challenge: Existing studies focus on specialized agents designed for particular tasks.
Approach: They propose to scale annotated interaction trajectories and fine-tune LLMs on AgentBank to get a series of agent models, Samoyed.
Outcome: The proposed model can scale to get generalized agent capabilities.
Learning Robust Representations for Continual Relation Extraction via Adversarial Class Augmentation (2022.emnlp-main)

Copied to clipboard

Challenge: Existing studies attribute catastrophic forgetting to the corruption of the learned representations as new relations come . Continual relation extraction models suffer from catastrophic forgetting when learning new relations .
Approach: They propose to use adversarial class augmentation mechanism to learn more precise representations . they propose to train the model on a sequence of tasks where two new relations are discovered .
Outcome: The proposed model improves on two popular benchmarks.
Beyond Single Frames: Can LMMs Comprehend Implicit Narratives in Comic Strip? (2025.findings-emnlp)

Copied to clipboard

Challenge: Large Multimodal Models have demonstrated strong performance on vision-language benchmarks, yet current evaluations focus on single-image reasoning.
Approach: STRIPCIPHER is a benchmark designed to evaluate model ability on understanding implicit narratives in silent comics.
Outcome: STRIPCIPHER is a high-quality, human-annotated dataset featuring fine-grained annotations and comprehensive coverage of varying difficulty levels.
More Tokens, Lower Precision: Towards the Optimal Token-Precision Trade-off in KV Cache Compression (2025.findings-emnlp)

Copied to clipboard

Challenge: storing more tokens in the KV cache at lower precision can enhance the long-context performance of large language models.
Approach: They propose a token-precision trade-off strategy to optimize KV cache compression . they also propose storing more tokens in the KV at lower precision .
Outcome: The proposed method achieves an optimal point within the Information Bottleneck compared to standalone KV pruning or KV quantization.
PACE: Prefix-Protected and Difficulty-Aware Compression for Efficient Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Existing LRMs often suffer from "overthinking" and excessively long reasoning traces . a dual-level framework for length compression of LRM is proposed .
Approach: They propose a framework for prefix-protected and difficulty-aware compression under hierarchical supervision.
Outcome: The proposed framework reduces token usage while improving accuracy on math benchmarks.
InfoCL: Alleviating Catastrophic Forgetting in Continual Text Classification from An Information Theoretic Perspective (2023.findings-emnlp)

Copied to clipboard

Challenge: Recent studies have identified the severe performance decrease on analogous classes as a key factor for catastrophic forgetting.
Approach: They propose a replay-based continual text classification method that uses fast-slow and current-past contrastive learning to perform mutual information maximization and better recover previously learned representations.
Outcome: The proposed method achieves state-of-the-art on three text classification tasks.
PLAWBENCH: A Rubric-Based Benchmark for Evaluating LLMs in Real-World Legal Practice (2026.acl-long)

Copied to clipboard

Challenge: Existing benchmarks for large language models (LLMs) are coarse, single-dimensional metrics and do not explicitly assess fine-grained legal reasoning.
Approach: They propose a Practical Law Benchmark to evaluate large language models in real-world legal practice scenarios.
Outcome: The proposed model is based on 850 questions and 13 scenarios with expert-designed evaluation rubrics.
FAER: Benchmarking VLMs for Failure-Aware Embodied Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Visual-language models (VLMs) are the core component of embodied agents in perceiving the environment and making decisions.
Approach: They propose a failure-aware benchmark to evaluate the performance of visual language models (VLMs) in long-horizon tasks.
Outcome: The proposed benchmark evaluates the performance of 16 widely utilized VLMs and 4 LLMs for FAER tasks.
GigaSpeech 2: An Evolving, Large-Scale and Multi-domain ASR Corpus for Low-Resource Languages with Automated Crawling, Transcription and Refinement (2025.acl-long)

Copied to clipboard

Challenge: GigaSpeech 2 is a large-scale, multi-domain, multilingual speech recognition corpus for low-resource languages.
Approach: They propose a large-scale, multi-domain, multilingual speech recognition corpus for low-resource languages and an automated pipeline for data crawling, transcription, and label refinement.
Outcome: The proposed corpus reduces the word error rate for Thai, Indonesian, and Vietnamese on a realistic YouTube test set by 25% to 40% compared to Whisper large-v3.
ParseJargon: Personalized Real-time Jargon Support in Online Meetings (2026.acl-demo)

Copied to clipboard

Challenge: Recent advances in speech-to-text technologies and large language models (LLMs) have the potential to overcome these limitations with automated, real-time jargon support.
Approach: They built an interactive LLM-powered system that provides real-time personalized jargon support tailored to users’ individual backgrounds in online meetings.
Outcome: The proposed system provides more precise jargon identification and enhanced participants’ comprehension, engagement, and appreciation of colleagues’ work.
Who Plays Which Role When? Communication Role Dynamics for Peer Recognition and Team Performance Prediction (2026.acl-long)

Copied to clipboard

Challenge: Prior work has modeled functional roles for meeting participants from simple speech features or used behavioral patterns to predict "latent" roles and team outcomes.
Approach: They operationalize a taxonomy of eight communication roles grounded in education literature and annotate a corpus of 6,307 Slack messages from 55 students across 18 teams.
Outcome: The proposed taxonomy outperforms lexical, conversational, and LLM-prompting baselines in predicting team performance after deliberation.
ConFiguRe: Exploring Discourse-level Chinese Figures of Speech (2022.coling-1)

Copied to clipboard

Challenge: Figures of speech often deviate from their literal meanings to express deeper semantic implications.
Approach: They propose a concept of figurative unit, which is the carrier of a figure, and build a Chinese corpus for Contextualized Figure Recognition.
Outcome: The proposed model is based on 12 types of figures commonly used in Chinese . it shows that the proposed tasks are challenging for existing models .
Glue pizza and eat rocks - Exploiting Vulnerabilities in Retrieval-Augmented Generative Models (2024.emnlp-main)

Copied to clipboard

Challenge: Retrieval-Augmented Generative (RAG) models enhance Large Language Models (LLMs) by integrating external knowledge bases.
Approach: They propose to exploit openness of RAG models by injecting deceptive content into the retrieval database, intentionally changing the model’s behavior.
Outcome: The proposed model can be exploited through crafted content uploads with access to the retriever.
Watch Every Step! LLM Agent Learning via Iterative Step-level Process Refinement (2024.emnlp-main)

Copied to clipboard

Challenge: Recent approaches to enhance agent performance focus on outcome rewards, which may lead to errors or suboptimal actions due to the absence of process supervision signals.
Approach: They propose a step-level framework that provides detailed step-by-step guidance to enhance agent training by using Monte Carlo methods.
Outcome: The proposed framework outperforms strong baselines on three tasks and shows that it is effective in augmenting efficiency and its applicability to diverse models.
ChatKBQA: A Generate-then-Retrieve Framework for Knowledge Base Question Answering with Fine-tuned Large Language Models (2024.findings-acl)

Copied to clipboard

Challenge: Existing KBQA methods address inefficient knowledge retrieval and semantic parsing errors.
Approach: They propose a generatethen-retrieve KBQA framework that generates logical form and replaces entities and relations with an unsupervised retrieval method to improve both generation and retrieval more directly.
Outcome: Experimental results show that ChatKBQA achieves new state-of-the-art performance on standard KBQA datasets, WebQSP, and CWQ.
Calibrating Factual Knowledge in Pretrained Language Models (2022.findings-emnlp)

Copied to clipboard

Challenge: Existing studies show that Pretrained Language Models can store factual knowledge, but facts stored in PLMs are not always correct.
Approach: They propose a lightweight method to calibrate factual knowledge in PLMs without re-training from scratch.
Outcome: The proposed method can be used to calibrate factual knowledge in PLMs without re-training from scratch.
MPO: Boosting LLM Agents with Meta Plan Optimization (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing methods for interactive planning tasks suffer from planning hallucinations and require retraining for each new agent.
Approach: They propose a framework that leverages explicit guidance through meta plans to assist agent planning and enables continuous optimization based on feedback from the agent’s task execution.
Outcome: The proposed framework outperforms existing baselines on two representative tasks and significantly improves task completion efficiency and generalization capabilities.
EcomScriptBench: A Multi-task Benchmark for E-commerce Script Planning via Step-wise Intention-Driven Product Association (2025.acl-long)

Copied to clipboard

Challenge: Goal-oriented script planning is used by humans to plan for typical activities . however, this capability remains underexplored due to several challenges .
Approach: They propose a framework that enables product-enriched scripts by associating products with each step based on the semantic similarity between the actions and their purchase intentions.
Outcome: The proposed framework can generate product-enriched scripts from 2.4 million scripts . human annotations are conducted to provide gold labels for a sampled subset .
Rationale-Enhanced Language Models are Better Continual Relation Learners (2023.emnlp-main)

Copied to clipboard

Challenge: Recent studies have found that catastrophic forgetting arises from the model’s lack of robustness against future analogous relations.
Approach: They propose a multi-task rationale tuning strategy to help the model learn current relations robustly and conduct contrastive rationale replay to further distinguish analogous relations.
Outcome: The proposed method outperforms the state-of-the-art models on two benchmarks.
The Good, The Bad, and The Greedy: Evaluation of LLMs Should Not Ignore Non-Determinism (2025.naacl-long)

Copied to clipboard

Challenge: Current evaluations of large language models (LLMs) focus on a single output per example, which limits our understanding of LLM performance variability in real-world applications.
Approach: They explore the performance differences between greedy decoding and sampling and identify benchmarks’ consistency regarding non-determinism and examine unique model behaviors.
Outcome: The proposed model outperforms sampling methods and greedy decoding outperformed other models.
Exploring Fine-Grained Human Motion Video Captioning (2025.coling-main)

Copied to clipboard

Challenge: Existing video captioning models fail to capture nuanced semantics of videos . existing models generate coarse descriptions of human motions, resulting in poor quality .
Approach: They construct a fine-grained human motion video captioning dataset named BoFiT and a model that generates fine-grain descriptions of human motions via prompting.
Outcome: The proposed model outperforms existing models on comprehensive metrics.
LongEmbed: Extending Embedding Models for Long Context Retrieval (2024.emnlp-main)

Copied to clipboard

Challenge: Existing embedding models support only 512 input tokens, hindering their application in scenarios requiring long inputs.
Approach: They evaluate the performance of existing embedding models by using a new benchmark and a training-free context window extension strategy.
Outcome: The proposed model extends the input window of existing models by several folds.
What Makes for Good Visual Instructions? Synthesizing Complex Visual Reasoning Instructions for Visual Instruction Tuning (2025.coling-main)

Copied to clipboard

Challenge: Experimental results show that visual instruction tuning improves performance of Multi-modal Large Language Models (MLLMs) to extend the application scope of Large Language Modells, a surge of work augments LLMs with vision encoders to endow the ability of multi-modal cognition and reasoning.
Approach: They propose a systematic approach to create high-quality visual reasoning instructions using a synthesize-complicate-reformulate paradigm.
Outcome: The proposed method improves performance of MLLMs by 27.86% and 27.60% on MME-Perception and MME Cognition.
Lock on Target! Precision Unlearning via Directional Control (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing methods for unlearning harmful, sensitive, or outdated knowledge suffer from two critical limitations: (1) collateral forgetting, where erasing target data inadvertently removes related but desirable knowledge, and (2) generality forgetting degrades the model’s general capabilities.
Approach: They propose a method that identifies and leverages a targeted "unlearning direction" in the model's parameter space and selectively updates along this direction.
Outcome: Experiments show that the proposed method achieves state-of-the-art unlearning precision while preserving both related knowledge and general capabilities.
CoUDA: Coherence Evaluation via Unified Data Augmentation (2024.naacl-long)

Copied to clipboard

Challenge: Existing data augmentations for coherence evaluation rely on heuristic rules and lack designing criteria.
Approach: They propose a data augmentation framework that breaks down coherence into global and local aspects and designs augmentation strategies for both aspects.
Outcome: The proposed framework surpasses recent models in scoring and ranking tasks with 233M parameters.
ISR: Self-Refining Referring Expressions for Entity Grounding (2025.acl-long)

Copied to clipboard

Challenge: Entity grounding is a crucial task in the construction of multimodal knowledge graphs.
Approach: They propose a novel scheme to enhance the multimodal large language model's capability to generate high quality REs for the given entities as explicit contextual clues.
Outcome: The proposed method surpasses other methods in entity grounding, highlighting its effectiveness, robustness and potential for broader applications.
SessionIntentBench: A Multi-task Inter-session Intention-shift Modeling Benchmark for E-commerce Customer Behavior Understanding (2026.findings-acl)

Copied to clipboard

Challenge: Existing models fail to capture and model customer intention effectively because of insufficient information exploitation and only apparent information like descriptions and titles are used.
Approach: They propose to exploit existing session data to capture and model intention in E-commerce product purchase sessions using a multimodal benchmark.
Outcome: The proposed framework can bridge the gap between intention understanding in simplified research cases like co-buy intention and more complex yet practical scenarios like session history.
FolkScope: Intention Knowledge Graph Construction for E-commerce Commonsense Discovery (2023.findings-acl)

Copied to clipboard

Challenge: Existing intention-based studies on recommendation tasks are limited and use models to implicitly model the intention memberships.
Approach: They propose a framework that leverages the generation power of large language models and human-in-the-loop annotation to semi-automatically construct the intention knowledge graph.
Outcome: The proposed framework can model e-commerce knowledge and have many potential applications.
EHRAG: Bridging Semantic Gaps in Lightweight GraphRAG via Hybrid Hypergraph Construction and Retrieval (2026.findings-acl)

Copied to clipboard

Challenge: Existing lightweight approaches to retrieval-augmented generation fail to capture latent semantic connections between disjoint entities.
Approach: They propose a lightweight RAG framework that constructs a hypergraph capturing both structure and semantic relationships using a hybrid structural-semantic retrieval mechanism.
Outcome: EHRAG outperforms state-of-the-art methods on four datasets while maintaining zero token consumption.
Trial and Error: Exploration-Based Trajectory Optimization of LLM Agents (2024.acl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have become integral components in various autonomous agent systems.
Approach: They propose an exploration-based trajectory optimization approach that allows agents to learn from their exploration failures.
Outcome: The proposed method outperforms baseline methods on three complex tasks by a large margin.

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