An N-Gram Framework for Sentiment and Emotion-Aware Word Association Games

Natural Language Processing (NLP) models have seen impressive advancements in understanding word associations. Still, limited attention has been given to user sentiment and emotions influencing these associations. In this paper, we explore the impact of both sentiment and emotion on the selection of words in an N-gram-based word association game. By integrating GloVe embeddings with SenticNet-based emotion classification and sentiment analysis from VADER, we evaluate how positive and negative sentiments, combined with intense emotions such as delight, ecstasy, terror, and loathing, affect word choices. Our findings suggest that user sentiment and emotion have a significant effect on word selection, with positive players associating positive adjectives with positive nouns, while negative players tend toward the opposite associations. This study highlights the necessity of considering sentiment and emotional intelligence in future NLP systems and presents new applications in areas such as AI-based gaming, behavioral analysis, and human-computer interaction.