The SGML language specifies the syntax of the elements (tags) in a document so that the information content can be retrieved.
The elements are defined for particular types (classes) of document in a Data type Definition (DTD). This together with the marked up (tagged) document forms an SGML application geared to solving a particular information retrieval problem.
Using SGML it is possible to retrieve the information on a element (field) by field basis (eg to generate reports).
HTML is a short, simple DTD aimed at basic layout and hyper-linking tasks in short documents.