ACL Findings2025
TDCSA: LLM-Guided Top-Down Approach for Robust Citation Sentiment Analysis
Fan Gao, Jieyang Peng, Xiaoming Tao, Youzheng Wang
Abstract
Citation Sentiment Analysis (CSA) plays a crucial role in understanding academic influence and knowledge diffusion. While pre-trained language models (PLMs) and large language models (LLMs) showed remarkable success in general sentiment analysis, they encounter specialized challenges in CSA due to the less significant and implicit sentiment expressions in academic writing, as well as complex sentiment transitions. In order to address the challenges, We propose TDCSA, a Top-Down framework that leverages LLMs' semantic understanding capabilities to enhance PLM-based CSA, which transforms the traditional bottomup feature engineering paradigm into a topdown architecture. Our framework consists of three key components: (1) a Dual LLM Feature Generation module for robust quadruple extraction, (2) a Multi-view Feature Representation mechanism for neutral citation processing, and (3) a Quad Feature Enhanced PLM. Experiments demonstrate that TDCSA 1 significantly outperforms existing methods, achieving state-of-the-art performance while maintaining robustness to quadruple quality variations.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- QLoRA: Efficient Finetuning of Quantized LLMs cites LoRA: Low-Rank Adaptation of Large Language Models
- QLoRA: Efficient Finetuning of Quantized LLMs cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
- MvP: Multi-view Prompting Improves Aspect Sentiment Tuple Prediction cites Aspect Sentiment Quad Prediction as Paraphrase Generation
- MvP: Multi-view Prompting Improves Aspect Sentiment Tuple Prediction cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Builds on8
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- LoRA: Low-Rank Adaptation of Large Language ModelsEdward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu et al.ICLR 2022 · 18,833 citations
- QLoRA: Efficient Finetuning of Quantized LLMsTim Dettmers, Artidoro Pagnoni, Ari Holtzman, Luke ZettlemoyerNeurIPS 2023 · 5,863 citations
- Aspect Sentiment Quad Prediction as Paraphrase GenerationWenxuan Zhang, Yang Deng, Xin Li, Yifei Yuan et al.EMNLP 2021 · 196 citations
- MvP: Multi-view Prompting Improves Aspect Sentiment Tuple PredictionZhibin Gou, Qingyan Guo, Yujiu YangACL 2023 · 48 citations
Related papers
- Adapting Pretrained Language Models for Citation Classification via Self-Supervised Contrastive LearningTong Li, Jiachuan Wang, Yongqi Zhang, Shuangyin Li et al.KDD 2025
- Sentiment-Aware Word and Sentence Level Pre-training for Sentiment AnalysisShuai Fan, Chen Lin, Haonan Li, Zhenghao Lin et al.EMNLP 2022 · 20 citations
- TimeMRA: LLM-Empowered Time Series Forecasting via Multi-Scale Retrieval-Augmented RepresentationsZongjiang Shang, Chengxi Jin, Binqing Wu, Dongliang Cui et al.ICML 2026
- DEQA: Descriptions Enhanced Question-Answering Framework for Multimodal Aspect-Based Sentiment AnalysisZhixin Han, Mengting Hu, Yinhao Bai, Xunzhi Wang et al.AAAI 2025 · 3 citations
- Systematic Task Exploration with LLMs: A Study in Citation Text GenerationFurkan Sahinuç, Ilia Kuznetsov, Yufang Hou, Iryna GurevychACL 2024
