DS@GT at eRisk 2025:从提示词到预测,基于对话代理评估和时间注意力模型的早期抑郁症检测基准测试
DS@GT at eRisk 2025: From prompts to predictions, benchmarking early depression detection with conversational agent based assessments and temporal attention models
AI总结:
DS@GT团队在eRisk 2025挑战赛中,利用提示词工程驱动多种LLMs进行基于BDI-II的对话式抑郁症检测,通过跨模型一致性评估,其最佳提交在官方排行榜排名第二。
摘要:
本工作笔记总结了 DS@GT 团队参与两项 eRisk 2025 挑战赛的情况。对于使用大语言模型(LLMs)进行对话式抑郁症检测的 Pilot Task,我们采用了提示词工程策略,由多种 LLMs 执行基于 BDI-II 的评估并生成结构化的 JSON 输出。由于缺乏真实标签,我们评估了跨模型一致性和内部一致性。我们的提示词设计方法使模型输出与 BDI-II 标准对齐,并能够分析影响症状预测的对话线索。我们表现最佳的提交在官方排行榜上排名第二,取得了 DCHR = 0.50、ADODL = 0.89 和 ASHR = 0.27 的成绩。
英文摘要:
This Working Note summarizes the participation of the DS@GT team in two eRisk 2025 challenges. For the Pilot Task on conversational depression detection with large language-models (LLMs), we adopted a prompt-engineering strategy in which diverse LLMs conducted BDI-II-based assessments and produced structured JSON outputs. Because ground-truth labels were unavailable, we evaluated cross-model agreement and internal consistency. Our prompt design methodology aligned model outputs with BDI-II criteria and enabled the analysis of conversational cues that influenced the prediction of symptoms. Our best submission, second on the official leaderboard, achieved DCHR = 0.50, ADODL = 0.89, and ASHR = 0.27.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。