Bing Search Engine: Thousands of Features in Search Ranking Models
On the search Algorithm sphere, Google declares of having about 200 ranking signals which stands contrasts to the declaration of having thousands of features in Bing’s search engine model by Frederic Dubut.
While Frederic’s comment on Twitter “The model has hundreds, if not thousands of features” stands contradictory to his own statement and seems to reveal an entirely different story.
Bing shares the belief that thousands of queries are recorded each day on the search engine. Sometimes the query is not clear and on the later half of that some queries with their nature of being the latest one doesn’t satisfy the user owing to the less available sources on the web. This fact is what led the Bing to divert its attention towards machine learning algorithm.
The concept of machine learning is identifying the patterns and deliver the results in accordance with the relevancy. The most relevant of the results filtered out by machine learning will bear the capability of viewing the unexplored sites, or it will discover fresh websites owing to giving most relevant results.
In short it means preparing a model which will scan out ideal SERPs in accordance with the relevance features. It will be a most appropriate way of creating a web ranking algorithm. It no more will require the expert’s suggestion for carefully engineering the problem.
It’s a grand move by Bing in stiff competition with Google, channelization of which is still a mystery.