针对心理防御机制分类中的类别不平衡问题，改进Qwen3-8B模型表现。
LinguIUTics at PsyDefDetect: Iterative Imbalance-Aware Fine-tuning of Qwen3-8B for Psychological Defense Mechanism Classification
- 采用分组分层交叉验证与轮转词汇增强，缓解少数类数据不足
- 通过后处理调整逻辑偏置并集成多模型，使关键类别F1达0.797
- 适合临床对话分析与罕见类识别任务的研究者参考
在对话文本中检测心理防御机制仍是临床NLP中的难题。针对PsyDefDetect 2026共享任务（九类话语分类，以宏平均F1为评估指标），我们团队LinguIUTics在官方正类排行榜上取得0.3917的宏平均F1，位列21支参赛队伍中的第4名，相比Ministral-8B基线（31.48宏平均F1）提升7.7个百分点（相对提升24.4%）。基于BERT的编码器和零样本大模型在稀有类别上表现不佳，主要因严重类别不平衡。因此我们采用QLoRA对Qwen3-8B进行迭代式不平衡感知微调。核心策略包括：分组分层交叉验证（防止信息泄露）、少数类轮转词汇增强、以及结合逻辑偏置调节与集成融合的后处理流程。三者协同显著缩小了验证集与排行榜差距，大幅提升少数类召回率，使关键类别“不确定”（第8级）的F1从接近零跃升至0.797。
原文摘要 · Abstract (English)
Detecting psychological defense mechanisms in conversational text remains a challenging clinical NLP problem. For the PsyDefDetect 2026 shared task (nine-class utterance classification evaluated via macro F1), our team LinguIUTics achieves a macro F1-score of 0.3917 on the official positive-class leaderboard, ranking 4th out of 21 registered teams and improving over the Ministral-8B task baseline (31.48 macro F1) by 7.7 absolute points (24.4 percent relative). BERT-family encoders and zero-shot LLMs proved ineffective on rare classes due to severe class imbalance, leading us to QLoRA fine-tuning of Qwen3-8B. We leverage three key strategies: grouped stratified cross-validation (preventing leakage), minority-class round-robin lexical augmentation, and a post-processing pipeline with logit bias tuning and ensemble blending. Together, these components close much of the validation-to-leaderboard gap and substantially improve minority-class recall, driving the critical "Unclear" class (Level 8) from near-zero performance to an F1 score of 0.797.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。