Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI