EMMR: Emotion-Mediated Multimodal Reasoning for Personality Assessment in Asynchronous Video Interviews
A new framework called EMMR (Emotion-Mediated Multimodal Reasoning) has been proposed for personality assessment in asynchronous video interviews (AVIs). The framework, detailed in a paper on arXiv (2608.07512), integrates emotion-related cues from visual and audio modalities into large language model (LLM) based personality assessment. Traditional text-centered methods often overlook non-verbal cues, which are crucial for understanding personality traits. EMMR extracts these cues and incorporates them as auxiliary social and behavioral evidence through structured reasoning. Experiments on two AVIs datasets, OPVA and AVI-6, demonstrate the effectiveness of the approach. The paper is categorized as a cross-type announcement and was published on arXiv.
Key facts
- EMMR is a two-stage framework for MLLMs-based personality assessment in AVIs.
- It extracts emotion-related cues from multimodal interview data.
- The framework incorporates these cues into personality assessment through structured reasoning.
- Experiments were conducted on OPVA and AVI-6 datasets.
- The paper is available on arXiv with ID 2608.07512.
- The announcement type is 'cross'.
- The approach addresses limitations of text-centered methods in personality assessment.
- Emotion-related cues provide social and affective evidence for personality traits.
Entities
Institutions
- arXiv