Trustworthiness of $\mathbb{X}$ Users: A One-Class Classification Approach