Search engines often provide an “advanced search” page, where a searcher can define search results they receive in many ways, beyond the simpler keyword search found on the front page search at those search engines.
For example, Yahoo’s advanced web page search lets searchers select a combination of different search limitations, such as:
- Different relationships between keywords in a search (e.g., “all of these words”, “the exact phrase”, “any of these words”, and “none of these words”),
- A time limitation on when the web page was last updated,
- A limit on what top-level domain names to search,
- A limit based on available legal rights,
- A file format limitation
- Country and language limitations.
If someone uses advanced search, they can significantly narrow the number of search results they receive, perhaps making it easier to find what they are looking for. But, most people don’t use the advanced search interface, and its many ways of limiting search results.
A personalized search method described in a Yahoo patent application published last week collects information about a searcher’s interests from their search history, their browsing history, and their interests listed in profiles from places like MySpace and other social networks.
It can use that information to limit the number of search results received for a query by possibly adding additional search query terms or using some of the other limitations available in the advanced search options.
Invented by Harshal D. Dedhia
Assigned to Yahoo
US Patent Application 20080109422
Published May 8, 2008
Filed: November 2, 2006
This personalized search uses information previously collected about a searcher’s interests and might be useful in narrowing down the scope of results returned to a searcher in response to a query.
This information can be collected through something like a toolbar or a browser helper program.
One way that the personalized search might work is to add one or more search terms to a searcher’s query if the searcher’s known prior interests are relevant to the query. It could also perform an advanced search, using a searcher’s known interests, some of Yahoo’s advanced search parameters, and the searcher’s original query terms.
For example, if the client has performed numerous searches for real estate in the San Francisco Bay area, the search system may select San Francisco Bay as a search parameter or search term automatically for future searches of the index for real estate.
While that could happen automatically, it’s also possible that a searcher might be prompted to see fewer results in certain instances.
In yet another embodiment, the client may be prompted for which results set to transmit and the prompt may include information about the limited search operation 110 or information providing a high-level comparison of the results set (e.g., “The full results of the search include over 1 billion pages, would you like to see the results of your search limited to the San Francisco Bay area?”, or “The full result set of your search `Iraq` and `war` is very large, would you like to limit this search to today’s news from major news sites?”).
The selection may then be made based on the client’s response to the prompt.
Where a Searcher’s Interest Information is Gathered
Some of a searcher’s interests may be gathered from a source unrelated to the search engine, such as a community web site that the search engine has access to. such as MySpace, Amazon.com, Barnes and Noble, and social networking sites.
Other interest information might be passively collected through the search engine and such things as a record of previous search requests.
Some information might be actively collected, such as geographic location and demographic information.
Relevance Analysis Between Searcher’s Interest and Searcher’s Query
The search term used by a searcher might be considered along with the searcher’s interest information, to see if there is some existing relationship between them, based upon relevance algorithms.
Search results from the original query, and results from limited searches from terms generated from the searcher’s interest information might be compared to determine how relevant they are to each other.
If a search query and searcher information fall within a subject category, e.g., a real estate query, a music query, local geographic query, a product query, a news query, etc. the search might be limited to certain parameters, such as limiting a search to a “news” category.
This patent application provides a pretty high-level view of how personalized search might work at Yahoo. What is interesting about it is how it can incorporate some of the advanced search options into how it functions.
Another interesting aspect of Yahoo’s approach is how it might use the profile information from eCommerce sites and social network sites to learn about a searcher’s interests.