Can LLM-Generated Textual Explanations Enhance Model Classification Performance? An Empirical Study
This study investigates the use of LLM-generated textual explanations to improve model performance across natural language inference tasks. It demonstrates competitive effectiveness compared to human-annotated explanations.
Advancing Data Equity: Practitioner Responsibility and Accountability in NLP Data Practices
This paper explores how NLP practitioners perceive and navigate issues of data equity, highlighting the tensions between commercial objectives and equity commitments.
LLMs for Law: Evaluating Legal-Specific LLMs on Contract Understanding
This evaluation reviews 10 legal-specific LLMs on English contract understanding tasks, showing significant performance differences compared to general-purpose models.
DAGR: Decomposition Augmented Graph Retrieval with LLMs
DAGR introduces a method that coherently connects complex QA with knowledge graphs, enhancing the efficiency of multi-hop reasoning in knowledge-intensive tasks.
HINTs: Sensemaking on large collections of documents with Hypergraph visualization and INTelligent agents
This paper presents HINTs, a visual analytics tool that integrates LLMs with document hypergraphs, enabling enhanced sensemaking on large datasets.
Data-Efficient Biomedical In-Context Learning: A Diversity-Enhanced Submodular Perspective
The proposed Dual-Div framework focuses on optimizing example selection in biomedical NLP tasks by enhancing diversity along with representativeness.
Evaluating Large Language Models as Expert Annotators
This paper evaluates the effectiveness of LLMs in replacing human expert annotators across finance, biomedicine, and law, using a multi-agent discussion framework.
APIO: Automatic Prompt Induction and Optimization for Grammatical Error Correction and Text Simplification
APIO introduces a method for inducing and optimizing prompts for grammatical error correction and text simplification without manual seed prompts.
Advancing Autonomous Incident Response: Leveraging LLMs and Cyber Threat Intelligence
This paper presents a framework for automating incident response by leveraging LLMs with cyber threat intelligence, aiming to reduce analyst workload and enhance response accuracy.
When Explainability Meets Privacy: An Investigation at the Intersection of Post-hoc Explainability and Differential Privacy in the Context of Natural Language Processing
The paper explores the intersection of explainability and privacy in NLP, investigating the challenges and potential for these two aspects to coexist.