Document similarity is a measure of how alike two text documents are, often used in information retrieval, clustering, and recommendation systems. It involves comparing documents based on their content, structure, or both, using techniques like vector space models and semantic analysis to quantify the degree of similarity.