Analysis, Evaluation, and Future Directions on Multimodal Deception Detection

Multimodal deception detection has received increasing attention from the scientific community in recent years, mainly due to growing ethical and security issues, as well as the growing use of digital media. A great number of deception detection methods have been proposed in several domains, such as political elections, security contexts, and job interviews. However, a systematic analysis of the current situation and the evaluation and future directions of deception detection based on cues coming from multiple modalities seems to be lacking. This paper, starting from a description of methods and metrics used for the analysis and evaluation of multimodal deception detection on video, provides a vision of future directions in this field. For the analysis, the PRISMA recommendations are followed, which allow the collection and synthesis of all the available research on the topic and the extraction of information on the multimodal features, the fusion methods, the classification approaches, the evaluation datasets, and metrics. The results of this analysis contribute to the assessment of the state of the art and the evaluation of evidence on important research questions in multimodal deceptive deception. Moreover, they provide guidance on future research in the field.