Emotional tendency analysis of game products based on improved deep reinforcement learning in MOBA environment

The need for player emotion in game product design and operation is growing in relevance given the explosive growth of the game sector. As a mainstream competitive genre, MOBA games involve regular player interactions and strong emotional fluctuations that quickly call for effective emotion tendency research techniques to improve user experience and product tuning impacts. In this study, we build an emotional model framework based on enhanced deep reinforcement learning (DRL) and propose a model entitled MOBA-SentDRL for MOBA environment. The results reveal that MOBA-SentDRL beats the conventional model in several criteria by building an actual MOBA game user sentiment dataset and assessing it in a range of comparison studies. The work motivates the future direction of multimodal and multi-strategy fusion for emotional computing and offers a practical method for the emotion interpretation and interaction of intelligent gaming goods.