North Carolina State University researchers have developed a search engine that is better at personalizing results for users, according to Phys.org. The researchers first wanted to solve the conundrum of the confusing results search engines sometimes display for specific searches. They wanted to create a search algorithm that uses “ambient query context” which is the process of looking at a user’s past queries and history to deliver the most optimized results.
For example, if someone searched, “mustang speed” and was previously looking up where the closest Ford dealership was, then the search engine can assume that the user is most likely looking for the speed of a Ford Mustang sports car, not the top speed of the animal mustang.
In addition to past history, the search engine also looks at what results a user clicks on to help provide better context for what a user is looking for. This means that if a user is looking for “mustang speed” but then begins clicking on results with the keyword of “trail rides” or “mustang ranch” then the search engine can take that context and use it to influence future query results based on what the user actually clicks on.
The Future of the Semantic Web
The point of the study was to prove that search engines on the Semantic Web should be moving toward contextual results, rather than providing results of heavily optimized pages based on keywords only. Many users are looking for search engines to automatically interpret their results.
However, the study argues that personalizing search results for each user may take up exponentially more computing power. Their succeeding versions of the search engine have been made more efficient than the first architecture model, but scalability may be an issue.