June 3, 2007

Google Shows More Fresh Results

New York Times has a long article about Google's search quality team and the way they constantly improve the ranking algorithms.

"Search over the last few years has moved from Give me what I typed to Give me what I want", says Amit Singhal from Google. His team tries to find patterns in the list of queries that return bad results, obtained from other Googlers or from users. Tweaking the ranking algorithm to favor some web pages in certain conditions is difficult because the results may change in unexpected ways.

One of the most important patterns from last year was that people expected to see fresh pages for queries related to recent events. For example, a search for "Google Finance" didn't return Google's financial site many days after the launch.

Mr. Singhal introduced the freshness problem, explaining that simply changing formulas to display more new pages results in lower-quality searches much of the time. He then unveiled his team's solution: a mathematical model that tries to determine when users want new information and when they don't. (And yes, like all Google initiatives, it had a name: QDF, for "query deserves freshness.") (...)

The QDF solution revolves around determining whether a topic is "hot." If news sites or blog posts are actively writing about a topic, the model figures that it is one for which users are more likely to want current information. The model also examines Google's own stream of billions of search queries, which Mr. Singhal believes is an even better monitor of global enthusiasm about a particular subject.

The visible part of QDF is the recently launched Hot Trends site, but this is just the tip of the iceberg. For queries related to things that are suddenly popular, Google's ranking algorithms are biased towards recent web pages. You may see results from Google News inside the search results pages or a blog search OneBox at the bottom of the page. Google also seems to be crawling pages at a much faster pace and not just for popular sites that are frequently updated, like they did before. I often see some of my posts in the search results hours after they're published.

7 comments:

  1. It is worth nothing that, in some ways, Google's search has gotten steadily worse over time. There are queries I could issue several months ago, especially site-specific ones, that no longer turn up anywhere ear as many hits.

    And people using Google to find older information will find it much harder to find if searches are biased in favor of recent updates.

    ReplyDelete
  2. Interesting. Can you give some examples of queries that return bad results?

    Google has always been biased towards the sites that passed the test of time, but these changes are applied only to a limited number of queries. So let's assume there's an earthquake in China and Google's top results for [earthquake China] are some pages that describe the most important earthquakes from China. But what happens if there's an earthquake right now? More people will search for [earthquake China] and they'll want to read news about this recent event.

    ReplyDelete
  3. I have to agree, Google's results are getting worse. In fact, I think it's because of this push for 'recent' results.

    I've often looked for statistics or quotes from the 90's so I can compare them to current situations in discussions. It's impossible to find them, as all I get is more recent stats and statements that don't help me in my search.

    IF they instead gave us a visible option to define date ranges for search results, it would be preferred in some situations.

    ReplyDelete
  4. "IF they instead gave us a visible option to define date ranges for search results, it would be preferred in some situations."

    Possibly the timeline thing in Google Labs(experimental search) could be an answer to this....

    ReplyDelete
  5. I don't care how fresh the search result is, I just want what I want. Many search results are duplicated. Could google choose the best one for us?

    ReplyDelete
  6. I have to agree that for some thing, google search is getting worse. I ofen try to look up details of old computer equipment via Google, since of the manufacturer site no londer exists. Often I did the information in the last page of search results, all the rest just being sites trying to sell me spares or replacements, and often morer looking further into them, they can't even do that!

    Thius is just one of my pet hates. However, I have noticed that now I'm using the personalized search, tthis is getting better.

    ReplyDelete
  7. Heh, I've noticed the faster indexing time. Last week I blogged about something I was researching in the morning. In the afternoon decided to do a bit more research on the same topic via Google, and found that my own page on the topic was the top hit!

    (And no, I didn't mis-spell it, which was my first guess.) :-)

    ReplyDelete