Search Engine Optimisation

Search Engine Marketing

Web Site Development

 

Google patent summary

Methods and systems for information extraction [US 2005/0131764 A1]

Methods documented in this Google patent describe methods for extracting shopping information from web based article.

The Google patent describes methods of how the Google search engine identifies web based documents which present details of items/services for sale and methods for extracting data elements, including price, Image , an SKU and version of the item. The document also describes how this information can be ranked and grouped for display to a client device in response to a client query.

A shopping document is first identified during or after a web crawl by the presence of at least one price representation and a shopping characteristic in a link or form element. The price representation is identified by the presence of a string with a currency identifier [£,$,€] and a decimal place or comma, followed by two digits. The shopping characteristic string can be present in the URL, parameter or value of an HTML element such as a <A>, <FORM>, <IMG> or <INPUT>. The string of a shopping characteristic would typically be 'Add to Basket', Add to Cart' or 'Buy now' etc.

The Google patent further describes the methods used to isolate the associated elements of the item for sale. The document describes how Google identifies potential price attributes by the format and proximity of keywords, such as 'sale', 'price', 'our price' or words to that effect and how Google dismisses inappropriate prices by the proximity of strings like 'was', 'save', 'from', 'starting at' or 'shipping'. Association of images is determined by the proximity, size, frequency and aspect ration of potential images. Images which occur frequently, are long or tall, small or large; are dismissed as logos, icons or other images and not used as a product image.

Conclusion

Although you may find some speculative documentation that claims a correlation between commercial web sites and Google natural SERPs, we find no evidence of this and advise no special formation of web pages to avoid a Google shopping document flag. If your document has items for sale; use shopping characteristics, price formation and image formats which clearly identify you sale items; if your document is not a shopping document but uses some characteristics which may be associated with this type of document then then be sure to identify those characteristics as none sale attributes.