Wednesday, November 5, 2008

Google Universal Search Patent Granted

Google was granted a patent today from the USPTO on Universal Search, which provides searchers with a mix of search results from different categories, such as news, images, advertisements, web pages, and kinds of results when they type in a search query

The original patent application was filed on December 31, 2003, and Google announced the introduction of Universal Search in May of 2007. The patent describes some different kinds of document categories that may be shown in search results, such as:

  • Sponsored links,
  • News documents,
  • Product documents,
  • Documents summarizing discussion groups,
  • Images,
  • General web documents, and;
  • Other document classifications

The Official Google Blog described a few more categories that could be shown to searchers in their announcement, Universal search: The best answer is still the best answer, including Maps, Books, Video, as well as additional contextual links to other categories of documents such as “blogs,” “books,” “groups,” and “code.”

Improving User Experience with Universal Search

The primary goal behind Universal Search, as noted in the patent, is to improve a user’s search experience without requiring them to have to choose a amongst different categories, such as images or news or web, before they send their query terms to the search engine.

Search engines have tried to provide access to different kinds of searches in the past, through the use of tabs or links above a search box that can lead to different kinds of results, such as news or images or products, but the patent’s authors tell us that “a large majority of users tend to ignore the category tabs, resulting in their search query being directed to the default category.”

Interface for a universal search engine
Invented by Bret Taylor, Marissa Mayer, and Orkut Buyukkokten
Assigned to Google
US Patent 7,447,678
Granted November 4, 2008
Filed: December 31, 2003

Choosing which Categories of Results to Show

Each of the different kinds of documents may be kept in separate databases, so that for instance, there’s a separate news category database, a separate product category data base, a separate image category database, and so on.

When someone performs a search, each of the databases may be searched to find the most relevant results in that database for the query terms that were entered by a searcher, and the results from each of those may be ranked.

The ranked results from each of the different data bases may then be compared to each other to see which provides the closest results to the search query.

For example, on a search for “buy running shoes, the results in the “products” category may be the most relevant to that search, and the ranking component may also look for terms like “buy” that indicate that a particular category may be related to a category like a “products” shopping category.

The patent also tells us that most searchers expect to see web page results when performing a search at a search engine, so a web page category will usually be the most prominent category for most searches.

Universal Search Interface

The interface shown for Universal Search in this patent is one where different groups of categories are shown in different segments of a page.

A more recent patent filing from Google has shown that the search engine has moved away from trying to group search results so strictly by category, blending different types of results together. I’ve written about that in How Google Universal Search and Blended Results May Work


The days of a search engine just providing a list of links in their search results are drawing further away as more kinds of content appear on the Web, and search engines are finding better ways of indexing that content.

News stories can present freshness in results, blog posts can provide unique perspectives, video can enable an alternative experience to reading, images can describe and tell a tale with one glance, and book results can lead to material that isn’t available online. Displaying results in these alternative categories of documents and others enables a richer experience for a searcher.

The pages of Google have remained very simple since the search engine was first introduced, and part of the reason for that was to make the site very fast, and easy to use even if you had a slow connection to the Web. As broadband access to the internet becomes more widespread, and as images and audio and video resources become more common, more colorful and complex results pages at search engines are more reasonable.

One of the challenges that site owners face is in presenting their audio and video and images and other non-text resources so that they can be found easily by people searching for what those site owners present on their pages. Universal and Blended search from the search engines provide ways for that material to be found by searchers.

This patent on Universal Search doesn’t tell us a lot that we haven’t already learned from using Universal Search for the last year and a half, but it does provide a few insights, such as a likelihood that most sets of seach results will always include Web pages because most searchers expect to see them, and that choice of different categories of results to present is tied most closely to how well the most relevant results in that category fit with the intent evidenced by a searcher’s query terms used during a search.

1 comment:

anab01 said...

I like the topic which has been discuss by you,i want to say thanks to you for sharing with me such a nice info.hpoe you'll soon give me more info regarding this topic,I'll waiting for that.
top bpo companies in india