How Google Could Categorize Sites as Low Quality Sites

On Tuesday, June 23, 2015, Barbara Starr and I are giving a two person presentation to the SEM San Diego Meetup, and Lotico Semantic Web San Diego Meetup Groups. This is titled ‘Ranking in Google from the dawn of the Knowledge Graph’.

During the preparation of the presentation, Barbara and I have glanced through a lot of patents. Quality Scores was one of the topics that were decided for the discussion. This is because one of the patents mention adding “Buy Now” buttons to paid search listings in search results is allowed to do so. This is only if the sites are able to show buy now buttons have a high enough Quality Score related to them.

During the preparation, Barbara showed another patent that emphasized on low quality scores. It highlights the way in which a site may lose traffic if ranking scores for links pointed to it are lesser than a particular level.

The focus of this patent is to provide a quality score to resources that link to a site. This is done by calculating the number of resources in every group, finding out a link quality score for the site by making use of the number of resources in each resource quality group. It could probably even verify that the link quality score is lower than a specific link quality score and categorize the site as a low quality site as the link quality score is lesser than the threshold link quality score. If the site is labeled as a low quality score site, it can bring about a reduction in the ranking score for that search result. This is calculated by an amount depending on the link quality score by using a formula that is listed in the patent.

The patent highlights some factors that could result in a lower quality score depending upon the links that direct to your site. Links to your site from boilerplate content are part of this. This also covers links from unnecessary material from the same sites that are pointed to your site. A good number of links like these could lead to a “low quality site” tag for your site.

At the moment, it is hard to guess if a patent like this one has started working or has any sort of effect. Reading articles like the one below makes you speculate if Google has actually bought into effect the details mentioned within this patent and other similar ones. The article is, Phantom 2 – Evaluating the Google Update that began On April 29, 2015

Linked to from Boilerplate
There are a good number of sites with content that could be regarded as boilerplate content. This could consist of footer content and sidebar content, which does not really add a lot of value to pages on a site and it is copied from one page to another. This could also cover copyright notices, links to things like sitemaps, and more. It can be seen that when a person links to another site from boilerplate sections of pages, there are chances of Google seeing this as a sign of low quality for the sites that are linked to.

Diversity filtering
Diversity filtering is also something the patent talks about. This is a process for removal of resources that offer unneeded information to the link quality engine. When your site offers a resource that links to a particular site and this done for a good number of pages, there is a possibility that Google may count only one of those links and abandon others. The filtered links against the site may be assumed by Google as a sign of low quality.

Advantages of the Patent
Recognize candidate resources that link to the site.

Remove all candidate resources that link to the site from a boilerplate section.

Abandon candidate resources by using a diversity filter.

Get resource quality scores for the left over resources.

Allocate the resources to resource quality groups.

Calculate the number of resources in every resource quality group.

Find out a link quality score for the site by making use of the number of resources in the resource quality groups.

Following the inventors of this patent could bring about benefits such as:
The link quality score for a site can be calculated by a search system by making use of distribution of resource quality scores of resources linking to the site

If the link quality score is lower than a standard score, the search system can consider the site as a low quality one.

The search can bring down the ranking score of a site considered as a low quality site while providing search results. This way, higher quality sites are offered to users in place of low quality sites

The patent is:
Categorizing sites as low quality sites
Publication number: US9002832 B1
Publication date: Apr 7, 2015
Filing date: Jun 4, 2012
Priority date: Jun 4, 2012
Inventors: Rajan Patel, Zhihuan Qiu, Chung Tin Kwok
Original Assignee: Google Inc.

Abstract
To improve search results, methods, systems, and apparatus, that cover computer programs programmed on a computer storage medium. A method usually has to do with receiving a resource quality score for each of a plurality of resources linking to a site. Every resource is delegated to one of a plurality of resource quality groups. Each resource quality group is associated with a series of resource quality scores. Every resource is allocated to the resource quality group linked with the range surrounding the resource quality score for the resource.

Calculation of the number of resources in each resources quality group is done. By making use of the number of resources in each resource quality group, a link quality score is established for the site. The site is deemed as a low quality site if the link quality score is lower than a threshold link quality score.

Take Aways
This is not the only patent from Google that has its focus on bringing down the rankings to sites depending on a low quality score for that site. Google’s Panda and Penguin updates also do pretty much the same.

fara