A document analysis apparatus 10 includes: a candidate generation unit 11 that, for each line included in a document that is a target of structural analysis, specifies another line in a parallel relationship with the line by performing extraction of a marker indicating a hierarchy, and generates a candidate for a hierarchical structure of the document that is the target based on the result of the specification of each line; and a candidate evaluation unit 12 that, if two or more candidates have been generated, performs evaluation on each candidate for the hierarchical structure, and selects one candidate for the hierarchical structure as the hierarchical structure of the document that is the target based on the evaluation result.