Following on from Google announcing page load speed to be a ranking factor, they have had a patent granted that was filed in 2004 for a new ranking signal, semantic closeness. This is when Google uses the HTML structure of a webpage to view how close words are together and determine a visual structure of the webpage that is semantically meaningful.

To do this Google will look at the titles and headings, list items (both ordered and unordered), nested table, divs (HTML code which determines where items are located on a page) and line breaks.

For example

Blackpool Tower Facts

• Height is 158 meters
• Weight is 2586 tonnes
• Made of 2493 tonnes of steel and 93 tonnes of cast iron
• 563 steps from the roof of the tower building to the flag pole
• Receives 650,000 visitors every year

By using the example above, semantic closeness can be defined as the distance between the header, in this case, “Blackpool Tower Facts” and the first word of the list entry. E.g. “height” or “weight”. Google will determine this visual structure to be and read it as “Blackpool Tower height” or “Blackpool Tower weight”. As both of these terms could easily be searched, Google will then rank this page as being relevant to these terms.

It also must be noted that words that appear in the same list item are also considered close. But words that appear in two different list items are considered to be further away, despite where the words fall in the list items.

This new patent does highlight how Google wants to be able to return searches the best, most relevant results when the search term is a question. E.g. “How much does Blackpool Tower weigh?”

Google Patent – Rankings based on semantic distance between terms in a document.