Affective-NLI: Towards Accurate and Interpretable Personality Recognition in Conversation

TL;DR

Affective-NLI tackles PRC by jointly leveraging affective cues from utterances and semantic descriptions of personality traits, reframing recognition as natural language inference. The method fine-tunes an emotion-recognition model to annotate utterances and uses descriptions of Big Five traits as hypotheses, enabling interpretable inferences about personality from conversation content. Across two datasets, Affective-NLI achieves consistent 6–7% improvements over strong baselines and demonstrates notable early-stage recognition in the Flow setting (0.5–0.6 accuracy with minimal dialog). This approach offers a practical, interpretable pathway for real-time personality-aware HCI systems, with potential extensions to capture trait correlations and multimodal signals.

Abstract

Personality Recognition in Conversation (PRC) aims to identify the personality traits of speakers through textual dialogue content. It is essential for providing personalized services in various applications of Human-Computer Interaction (HCI), such as AI-based mental therapy and companion robots for the elderly. Most recent studies analyze the dialog content for personality classification yet overlook two major concerns that hinder their performance. First, crucial implicit factors contained in conversation, such as emotions that reflect the speakers' personalities are ignored. Second, only focusing on the input dialog content disregards the semantic understanding of personality itself, which reduces the interpretability of the results. In this paper, we propose Affective Natural Language Inference (Affective-NLI) for accurate and interpretable PRC. To utilize affectivity within dialog content for accurate personality recognition, we fine-tuned a pre-trained language model specifically for emotion recognition in conversations, facilitating real-time affective annotations for utterances. For interpretability of recognition results, we formulate personality recognition as an NLI problem by determining whether the textual description of personality labels is entailed by the dialog content. Extensive experiments on two daily conversation datasets suggest that Affective-NLI significantly outperforms (by 6%-7%) state-of-the-art approaches. Additionally, our Flow experiment demonstrates that Affective-NLI can accurately recognize the speaker's personality in the early stages of conversations by surpassing state-of-the-art methods with 22%-34%.

Figures (4)

• Figure 1: When the robot lacks knowledge about Mrs. Thompson's personality, its comfort is generic and straightforward. However, if the robot deduces that Mrs. Thompson exhibits Neuroticism based on the dialog content, it can tailor its suggestions to offer more personalized and comforting advice.
• Figure 2: An overview of Affective-NLI with the first four utterances in Figure \ref{['toy_example']}. The upper-left part illustrates the affective dialog content construction. The lower-left part shows the positive and negative label descriptions of Neuroticism. The right part shows (1) how we construct NLI prompting samples with the affective dialog content and both positive and negative personality label descriptions and (2) how we combine both NLI results for the final personality recognition. Noting that the two different inputs to the same pre-trained language model.
• Figure 3: Personality recognition accuracy comparison among the sub-models in the ablation study.
• Figure 4: The personality recognition accuracy comparison in the Flow experiment. The transparent error bars indicate the lower and upper bounds of the current accuracies.