Despite significant advancements in the performance and capabilities of data-driven machine learning systems, establishing trust remains a profound challenge. Complex trust paradoxes, such as balancing transparency with performance, and personalization with privacy, complicate the establishment of trust. This paper builds on the main foundations of trust in technology from a systematically review of literature, focusing on the interplay between trust components and the tensions among competing favorable features and ethical principles. We then introduce a new trust ontology that captures the key elements of trust and their relationships. This ontology highlights the dynamics of trust, emphasizing the conflicting relationships between favorable features and how they influence trust between humans and machine learning systems.