Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SILC-EFSA: Self-aware In-context Learning Correction for Entity-level Financial Sentiment Analysis

Senbin Zhu, Chenyuan He, Hongde Liu, Pengcheng Dong, Hanjie Zhao, Yuchen Yan, Yuxiang Jia, Hongying Zan, Min Peng

TL;DR

A novel two-stage sentiment analysis approach called Self-aware In-context Learning Correction (SILC), which involves fine-tuning a base large language model to generate pseudo-labeled data specific to the authors' task.

Abstract

In recent years, fine-grained sentiment analysis in finance has gained significant attention, but the scarcity of entity-level datasets remains a key challenge. To address this, we have constructed the largest English and Chinese financial entity-level sentiment analysis datasets to date. Building on this foundation, we propose a novel two-stage sentiment analysis approach called Self-aware In-context Learning Correction (SILC). The first stage involves fine-tuning a base large language model to generate pseudo-labeled data specific to our task. In the second stage, we train a correction model using a GNN-based example retriever, which is informed by the pseudo-labeled data. This two-stage strategy has allowed us to achieve state-of-the-art performance on the newly constructed datasets, advancing the field of financial sentiment analysis. In a case study, we demonstrate the enhanced practical utility of our data and methods in monitoring the cryptocurrency market. Our datasets and code are available at https://github.com/NLP-Bin/SILC-EFSA.

SILC-EFSA: Self-aware In-context Learning Correction for Entity-level Financial Sentiment Analysis

TL;DR

A novel two-stage sentiment analysis approach called Self-aware In-context Learning Correction (SILC), which involves fine-tuning a base large language model to generate pseudo-labeled data specific to the authors' task.

Abstract

In recent years, fine-grained sentiment analysis in finance has gained significant attention, but the scarcity of entity-level datasets remains a key challenge. To address this, we have constructed the largest English and Chinese financial entity-level sentiment analysis datasets to date. Building on this foundation, we propose a novel two-stage sentiment analysis approach called Self-aware In-context Learning Correction (SILC). The first stage involves fine-tuning a base large language model to generate pseudo-labeled data specific to our task. In the second stage, we train a correction model using a GNN-based example retriever, which is informed by the pseudo-labeled data. This two-stage strategy has allowed us to achieve state-of-the-art performance on the newly constructed datasets, advancing the field of financial sentiment analysis. In a case study, we demonstrate the enhanced practical utility of our data and methods in monitoring the cryptocurrency market. Our datasets and code are available at https://github.com/NLP-Bin/SILC-EFSA.
Paper Structure (26 sections, 2 equations, 7 figures, 8 tables)

This paper contains 26 sections, 2 equations, 7 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Entity-level Sentiment Analysis of Financial Texts
  4. LLMs in Finance
  5. Datasets Construction
  6. English Dataset
  7. Chinese Dataset
  8. Methodology
  9. Stage 1: Initial Response Generation
  10. Stage 2: Self-correction Steps
  11. Experiments
  12. Baselines
  13. Experimental Settings
  14. Overall Performance
  15. Ablation Study
  16. ...and 11 more sections

Figures (7)

  • Figure 1: Examples of financial entity-level sentiment analysis data in English and Chinese. In the task, the objective is to identify financial entities within the text and analyze their sentiment within the context. Specifically, this involves annotating each entity along with its position (span) in the text and determining its sentiment polarity (e.g., positive, negative, or neutral).
  • Figure 2: Overview of our framework.
  • Figure 3: Performance impact of different numbers of in-context examples on the FinEntity dataset.
  • Figure 4: Impact of the ratio of correct data samples retained.
  • Figure 5: The English and Chinese instruction template for ChatGPT.
  • ...and 2 more figures