Research map / Large language models / Reasoning and chain of thought
Exploring Selective Avoidance for Online User Behavior Analysis: A Forest of Thought Explanation
Xiaohua Wu, Lin Li, Kaize Shi, Xiaohui Tao, Jianwei Zhang, Yuefeng Li
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
The paper addresses non-linear response behaviors (conditional branching, selective avoidance) in online user-generated content and proposes a Forest of Thought explanation approach to improve LLM reasoning trustworthiness for user behavior analysis.
Problem
Online UGC response behaviors show non-linear patterns such as conditional branching and selective avoidance, which unidirectional left-to-right LLM inference may not capture, undermining trustworthy reasoning.
Approach
Based on the title, a "Forest of Thought" explanation framework is proposed to model non-linear reasoning over selective avoidance in user behavior; the abstract does not give further details of the mechanism.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace FoT with FoTE (Shapley thought selection plus sampled selective avoidance) | FoT · Social Survey, Llama3-8B | success rate: 0.67 -> 0.72 | Improved |
| replace FoT with FoTE | FoT · Happy Moments, DeepSeek-R1-8B | success rate: 0.65 -> 0.74 | Improved |
| replace FoT with FoTE | FoT · Student Survey, DeepSeek-R1-8B | success rate: 0.66 -> 0.68 | Improved |
| replace FoT with FoTE | FoT · Student Survey, Qwen2.5-7B | success rate: 0.82 -> 0.84 | Improved |
| add explicit chain-of-thought reasoning over I/O prompting | Qwen2.5-7B · Social Survey | success rate: 0.20 -> 0.45 | Improved |
| replace CoT with SC-CoT, ToT, FoT variants | DeepSeek-R1-8B · Social Survey | success rate: 0.36 -> 0.53 | Improved |
| increase number of sub-trees to 64 (ablation) | FoTE · Social/Student Survey, Llama3-8B | success rate | Improved |
| increase number of sub-trees beyond 64 (ablation) | FoTE · Social/Student Survey, Llama3-8B | success rate | No clear effect |
| apply FoTE to social media tasks vs survey tasks (ablation) | FoTE · Depression Post vs surveys, Llama3-8B | success rate: 0.84 vs 0.8 peak | Mixed |
Closest papers by meaning
- Failing to Falsify: Evaluating and Mitigating Confirmation Bias in Language Models (NeurIPS 2026 · similarity 0.85)
- Thought Branches: Interpreting LLM Reasoning Requires Resampling (ICLR 2026 · similarity 0.85)
- Measuring the Faithfulness of Thinking Drafts in Large Reasoning Models (NeurIPS 2025 · similarity 0.85)
- Theory of Mind and Persuasion Beyond Conversation: Assessing the Capacity of LLMs to Induce Belief States via Planning and Action (NeurIPS 2026 · similarity 0.85)
- Look-Ahead Reasoning on Learning Platforms (NeurIPS 2025 · similarity 0.85)
- LLM Strategic Reasoning: Agentic Study through Behavioral Game Theory (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
