Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ECon: On the Detection and Resolution of Evidence Conflicts

Cheng Jiayang, Chunkit Chan, Qianqian Zhuang, Lin Qiu, Tianhang Zhang, Tengxiao Liu, Yangqiu Song, Yue Zhang, Pengfei Liu, Zheng Zhang

TL;DR

A method for generating diverse, validated evidence conflicts to simulate real-world misinformation scenarios and finds NLI and LLM models exhibit high precision in detecting answer conflicts, though weaker models suffer from low recall and stronger models like GPT-4 show robust performance.

Abstract

The rise of large language models (LLMs) has significantly influenced the quality of information in decision-making systems, leading to the prevalence of AI-generated content and challenges in detecting misinformation and managing conflicting information, or "inter-evidence conflicts." This study introduces a method for generating diverse, validated evidence conflicts to simulate real-world misinformation scenarios. We evaluate conflict detection methods, including Natural Language Inference (NLI) models, factual consistency (FC) models, and LLMs, on these conflicts (RQ1) and analyze LLMs' conflict resolution behaviors (RQ2). Our key findings include: (1) NLI and LLM models exhibit high precision in detecting answer conflicts, though weaker models suffer from low recall; (2) FC models struggle with lexically similar answer conflicts, while NLI and LLM models handle these better; and (3) stronger models like GPT-4 show robust performance, especially with nuanced conflicts. For conflict resolution, LLMs often favor one piece of conflicting evidence without justification and rely on internal knowledge if they have prior beliefs.

ECon: On the Detection and Resolution of Evidence Conflicts

TL;DR

A method for generating diverse, validated evidence conflicts to simulate real-world misinformation scenarios and finds NLI and LLM models exhibit high precision in detecting answer conflicts, though weaker models suffer from low recall and stronger models like GPT-4 show robust performance.

Abstract

The rise of large language models (LLMs) has significantly influenced the quality of information in decision-making systems, leading to the prevalence of AI-generated content and challenges in detecting misinformation and managing conflicting information, or "inter-evidence conflicts." This study introduces a method for generating diverse, validated evidence conflicts to simulate real-world misinformation scenarios. We evaluate conflict detection methods, including Natural Language Inference (NLI) models, factual consistency (FC) models, and LLMs, on these conflicts (RQ1) and analyze LLMs' conflict resolution behaviors (RQ2). Our key findings include: (1) NLI and LLM models exhibit high precision in detecting answer conflicts, though weaker models suffer from low recall; (2) FC models struggle with lexically similar answer conflicts, while NLI and LLM models handle these better; and (3) stronger models like GPT-4 show robust performance, especially with nuanced conflicts. For conflict resolution, LLMs often favor one piece of conflicting evidence without justification and rely on internal knowledge if they have prior beliefs.
Paper Structure (42 sections, 6 equations, 17 figures, 23 tables)

This paper contains 42 sections, 6 equations, 17 figures, 23 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Answer and factoid conflicts
  4. Conflict detection
  5. Conflict resolution
  6. Conflict detection
  7. Answer conflicts detection
  8. Evaluation setup
  9. Main results and analysis
  10. Detection under pollution attacks
  11. Factoid conflicts detection
  12. Evaluation setup
  13. Analysis on data
  14. Results and analysis
  15. Conflict resolution
  16. ...and 27 more sections

Figures (17)

  • Figure 1: Generating evidence pairs with answer conflicts. For each question and its ground-truth answers, alternative answers are generated (shown in red boxes). Subsequently, a piece of supporting evidence is generated for each answer, which is validated by a checker to ensure quality.
  • Figure 2: Type distributions of the answer and factoid conflicts.
  • Figure 3: Generating evidence pairs with factoid conflicts.
  • Figure 4: Distribution of conflict resolution behaviors.
  • Figure 5: Proportions of factoid conflict resolution behaviors, stratified by annotated intensity of conflicts.
  • ...and 12 more figures