To clarify, I am referring to the information present on the webpage which the users using the website are already consuming. The page already has a structure, thats how browser is able to display it to end user. Its just not tailored explicitly to a agent.

Lets leave the image vs text from the discussion from now, its a altogether different challenge.