This week has seen some chatter about a recent Google update, with the New York Times reporting a significant impact on traffic. While this change could have been the result of BERT, a new algorithm Google recently implemented, it very well could be something else.
In his post, ‘Evolution of Google’s News Ranking Algorithm,’ Bill Slawski suggests it’s possible that the algorithm behind how news articles rank at Google has changed. Read Bill’s article if you’re interested in learning more about this patent. He is the foremost expert on Google patents and offers great insight and explanation as to their function.
A Google Patent about how news articles are ranked by Google was updated this week, and in this case it suggests how entities in those documents can have an impact on ranking.Bill Slawski, Editor, SEO by the Sea
First filed in 2003, it has gone through a number of changes (continuation patents). Originally, the ranking of news sources had a great influence on the ranking of news articles. That meant large news organizations had an easier time ranking news articles at the expense of smaller organizations, regardless of the quality of their output.
The way news articles are ranked has improved over the years, with greater emphasis being placed on the actual content. Here is Google’s latest patent claim, granted October 29, 2019:
A method for ranking results, comprising: receiving a list of objects; identifying a first object in the list and a first source with which the first object is associated; identifying a second object in the list and a second source with which the second object is associated; determining a quantity of named entities that (i) occur in the first object that is associated with the first source, and (ii) do not occur in objects that are identified as sharing a same cluster with the first object but that are associated with one or more sources other than the first source; computing, based at least on the quantity of named entities that (i) occur in the first object that is associated with the first source, and (ii) do not occur in objects that are identified as sharing a same cluster with the first object but that are associated with one or more sources other than the first source, a first quality value of the first source using a first metric, wherein a named entity corresponds to a person, place, or organization; computing a second quality value of the second source using a second metric that is different from the first metric; and ranking the list of objects based on the first quality value and the second quality value.Google Patent: Systems and methods for improving the ranking of news articles
In other words, Google groups together news articles that talk about a specific topic. It creates a model or representation of what that topic is and then evaluates all those pages against the model. Pages that cover more of the topics, entities in Google parlance, in the model may rank higher.
An example should help make this a little clearer. Let’s use MarketMuse Newsroom to look at a current hot topic “Trump Impeachment.”
Here’s a closeup showing a partial list of topics in the topic model created by MarketMuse for the topic “Trump impeachment.”
These are the entities one should include in a news article covering this topic. So, if you’re writing a news article about the Trump impeachment, you should mention:
- vice president
- the justice department
- national intelligence
- the attorney general
- and other related topics in the list
By including these entities in your article, you provide comprehensive coverage of the news event. Not only is it good for your audience, it can help you rank better. According to the latest continuation of their patent, Google is evaluating news articles in this manner.
While we don’t know for sure if there was an update to the Google News algorithm, one thing is certain. In light of this latest Google patent, astute publishers will appreciate the benefit in making news articles more informative and comprehensive, thereby improving their ranking potential. MarketMuse Newsroom can help you do this at scale.
Written by Stephen Jeske