The Atom spec says that any title marked with a type of "xhtml" should be contained in a div element[1] so we need to use the full XML text when extracting the text. [1] https://www.rfc-editor.org/rfc/rfc4287#section-3.1