Using AI and Sentiment Analysis to Improve Word Association Games

What is it about?

This work started from a simple card game idea, like Apples to Apples, where you see an adjective like “excited” and try to pick the noun card that fits it best. We asked how an AI player could make that choice in a way that actually reflects mood and personality, not just random guessing. Our model looks at how closely words are related in meaning, whether they feel positive, negative, or neutral, and what kind of emotion they carry, like delight, serenity, terror, or loathing. To give the AI a sense of different “types” of players, we build small profiles from real tweets, for example, users who usually sound positive or usually sound negative. Then the AI uses those profiles to decide which card it would play for a given adjective. Overall, the project shows how combining AI, sentiment analysis, and basic emotion information can make a simple word association game feel more personal and human.

Featured Image

Photo by srihari kapu on Unsplash

Photo by srihari kapu on Unsplash

Why is it important?

We think this matters because a lot of AI systems still ignore how language actually feels. Right now, AI is being added to games and apps everywhere, so it needs to handle emotion more carefully. Most models score words only on raw similarity, even when the task is really about mood and tone. By building an emotion and sentiment-aware judge for a simple word game, we show a small, concrete way to make AI responses line up better with human intuition. The same ideas could carry over to things like chatbots, learning tools, or mental health apps, where it is important that the system not only understands what you say, but also how you sound.

Perspectives