Scholarlink

Identifying truthful answers in web-based questionnaires can drastically increase the validity of the collected data. Approaches based on the cognitive load of the person giving the answers have been successfully applied. Often, they rely on measuring the cognitive load with one modality (e.g., changes in the pupil diameter). In this paper, we present a bimodal approach that combines two modalities (i.e., the pupil diameter, and mouse movements). It automatically generates truth scores and weighting factors of the different modalities and produces human-readable graphs that allow study administrators to understand the background of the produced scores and weights.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Interpretable Truth Detection: Interpretable Bimodal Within-Subject Truth and Deception Detection Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Moritz Maleck,
  • Tom Gross

摘要

Identifying truthful answers in web-based questionnaires can drastically increase the validity of the collected data. Approaches based on the cognitive load of the person giving the answers have been successfully applied. Often, they rely on measuring the cognitive load with one modality (e.g., changes in the pupil diameter). In this paper, we present a bimodal approach that combines two modalities (i.e., the pupil diameter, and mouse movements). It automatically generates truth scores and weighting factors of the different modalities and produces human-readable graphs that allow study administrators to understand the background of the produced scores and weights.