Modern national security environments require advanced tools that can adapt dynamically to real-time data and human-centered inputs for decision makers to perform at a high-level. This paper introduces an approach which combines Multi-Objective Reinforcement Learning (MORL) with sentiment-driven dynamic weighting to improve decision-making using publicly available data from the Armed Conflict Location and Event Data Project (ACLED). The model adjusts its reward calculation based on human-centered sentiment inputs and analysis which allows for an adaptable policy decision in response to complex events like protests, battles, or violence against civilians. The outlined approach expands on transitional decision-making frameworks by including variable weight adjustments for actions like increasing aid, military presence, or deploying diplomatic resources. A Q-learning-based model shows how improved policy outcomes with sentiment-driven variables can show how much variability and adaptability can be applied across different scenarios. This approach offers a flexible and effective model to integrate real-time sentiment analysis for decision making.