Yahoo’s Social Search Answer to Google’s Aardvark?

In February of 2010, Google purchased a social Q&A site, Aardvark. It seems like a great match, for a couple of reasons. One is that a paper from Aardvark that attracted a lot of attention, The Anatomy of a Large-Scale Social Search Engine (pdf), written by Damon Horowitz and former Googler Sepandar D. Kamvar, was admitted by its authors to be inspired by one of the early Google papers, The Anatomy of a Large-Scale Hypertextual Web Search Engine. Another is that Aardvark’s founders and senior team members include a number of former Google (and Yahoo) employees.

Instead of looking for web pages that might answer your questions, Aardvark enables you to ask questions of people in your expanded social network (and beyond), and to identify topics that you might be interested in answering. While there are a number of Question and Answer type sites on the Web, such as Yahoo Answers, those don’t send out questions quickly to people who might be able to provide an answer, but rather rely upon people maybe happening upon your question.

A Yahoo patent application published this week explores a “communal search” system where someone might get real time responses to questions from people who might know the answers. People chosen to respond to questions might be selected based upon their location, activities they participate in, or some relationship to a location or time and the query. This system may also attempt to automatically answer queries based on previous questions and answers from others who have used the system.

The patent filing is:

System and Method for Communal Search
Invented by Chris Kalaboukis, Elizabeth F. Churchill, and Athellina Athsani
Assigned to Yahoo
US Patent Application 20100198869
Published August 5, 2010
Filed February 2, 2009


A system and method which may allow a requester to post a search query and get time-sensitive responses from other users who are best suited to respond to the search query. The system may find out such best suited responders based on their location, activity and time affinity/proximity to the search query, and forward the search query to them.

The system may then return the responses to the requester. The system may also be able to automatically answer search queries based on users’ historical query-response activities. By matching a search query with users knowledgeable about it, the invention may help a requester to get timely and accurate responses to his search query.

Someone asking a question might enter into the communal search system information such as:

  • A query type
  • A query title
  • A query description
  • Activity information
  • A start and end time, and maybe
  • Information about who he or see wants the query directed to.

If the query involves a location, such as “where’s a good place to park near the Washington Monument,”another interface might be used that includes location.

This system might include a tracking server that can get current status information from members of the community who might answer questions, such as their current location (via something like GPS or some other positioning system), or if they are involved in some kind of activity such as traveling at a speed that might indicate they are driving or in public transporation, or in other manners, as described in the following paragraph from the patent filing:

The tracking server may track or infer a user’s activities by monitoring his movements. When the user is moving at a certain speed or following a certain pattern, the tracking server may decide that the user is dancing. In one embodiment, the tracking server may determine the user’s activities by where the user is, e.g., whether he is at a museum, a movie theater, a gym, or a stadium.

In one embodiment, the tracking server may determine the user’s activities by social context. For example, if the user is in a big crowd of pop-music lovers, the tracking server may determine that the user is at a pop-music concert.

The tracking server may infer a user’s activity information from information on the Internet, e.g., the user’s on-line calendar, his statuses on instant messaging or status application such as or

This search system might also gather future status information by doing things like looking at planned calendar events a person is going to participate in.

It might also look at someone’s online activities such as:

  • Web pages browsed
  • Time spent browsing
  • Online shopping records
  • Online services registered for
  • Web content downloaded
  • Web content uploaded
  • Flicker tags – including geo-tags
  • Instant Messaging
  • Search queries input or responded to
  • Online calendar, blog, tweets, Facebook status updates, etc.

Other information that could be collected about a person who might participate in this communal search may include such things as:

  • Whether he or she has been to a searched location before, his affinity to that location, and/or his traveling profile.
  • A frequency profile (e.g., a user goes to his gym at least two times a week).
  • A routine schedule (e.g., a user is at work from 9 am to 5 pm, Monday to Friday).
  • An affinity for a particular type of activity, such as a fondness for visiting sushi restaurants
  • An expertise in a certain field, such as a blogger who often publishes reviews about cameras

Someone participating can also fill out information on a profile that includes things such as:

  • Area of expertise
  • Hobbies
  • Favorite singers
  • Sports he or she likes
  • Places familiar with
  • Schedule
  • Time period available to respond to search queries
  • Locations familiar with
  • Favorite restaurants, shops, and movie theatres
  • Activities he or she likes and dislikes
  • Information about friends/contacts/etc. in social networks like Facebook


We don’t know if Yahoo will ever develop this communal search system, if they’ve been hard at work developing it, or if it just seemed like a good idea to save it to paper.

There are a number of similarities with Aardvark, but Yahoo’s system seems much more intrusive in terms of what kinds of information they might collect, and how they might track the location and activities of people involved. The Yahoo patent filing does mention in one spot that it might limit questions and answers to people who may know each other, possibly by virtue of sharing a connection on a social network. And it’s possible that if this system is developed, the final product may not include a number of the features described in the patent filing, and may include others.

Is there a need for this kind of communal search, with services such as twitter and facebook and others around?

I hadn’t tried Aardvark before, so I signed up about an hour ago, and asked “Where is the best place for pizza in Northern Virginia.” A guy in Santa Monica, California, responded to my question with a suggestion for a place in about 20 minutes. Maybe there is some value. I’ll know after I try the pizza.


Author: Bill Slawski

Share This Post On