This thesis harnesses recent NLP advancements to create sentiment indices that capture public and investor moods through nearly half a million news articles. The novel sentiment measures align closely with established economic indicators, offering actionable insights for investors, businesses, policymakers, and regulators. With minimal cost and near-continuous updates, the approach outlined in this research provides a timely, data-driven alternative to traditional methods, empowering a more responsive, human-centered economic outlook. By rapidly identifying sentiment trends, these measures support informed, sustainable decisions, helping stakeholders understand and adapt to the emotional and social dynamics that shape markets, and possibly other areas.