Yahoo! has introduced a new patent that may help to create more personalized search results with a process called trend analysis.
Shyam Kapur of Yahoo! applied for the patent on May 12th, citing the need for “systems and methods for processing search requeststo provide a more sophisticated understanding of the information being sought.”
The implication of trend analysis as it applies to Yahoo! is that the system allows for significant and specific user information gathering and analysis, to better tailor search results, to give better demographic information, and provide information on consumer’s past search habits.
The trend analysis module looks for trends along different information parameters such as time of the queries (day, week, et cetera), where the user is searching from, what information is being sought, what they’ve looked for in the past, how older the user is, and how much money they make.
For an example of the future of these capabilities, a user who is a 35 year-old computer programmer making $50,000 a year in Peoria, Illinois will receive a different search result list when he types in “java,” than a 22 year-old coffee house frequenting art student in southern California.
THE CONFLICT BETWEEN PEOPLE THINK AND COMPUTER THINK
The patent application addresses how search engines and algorithms have a general inability to recognize contexts and conceptual word linkage. For example, a query for “New York City” is different than a query for “New York City Law Enforcement.”
Search engines as they are can only recognize that “New York City” is mentioned in both results, but the relevance is lost. The newly patented process will also build associations between “law enforcement” and “police,” allowing for nuance and “relatedness” results.
An excerpt from the patent application:
Current technologies at any of the major search providers, e.g., MSN, Google or any other major search engine site, do not understand queries the same way that human beings create them. For instance, existing search engines generally search for the exact words or phrases the user entered, not for the underlying natural concepts or related concepts the user actually had in mind. This is perhaps the most important reason that prevents search providers from identifying a user’s intent and providing optimal search results and content.
The trend analysis system aims to mirror the way people conceptualize search terms.
Another example of personalization is for general contexts. A 21-year-old enters the search term “singers” and the system could be more likely to bring up “Avril Lavigne” than “Celine Dion” who would be more popular among older users.
It is warned that the results may not be 100% accurate, as everything depends upon user behavior.
“It should be understood that these inferences are not guaranteed to be accurate; they reflect how the user behaves and not necessarily who the user is. Any inferences made about a particular user can be used to tailor responses to further queries entered by that user, again using the trend data as one indication of likely user intent. Thus, trend data may be used to customize the response of a search server to the particular user who enters a query,” according to the patent application.
