Comments on: Wikia Search: Social Search Is Blooming http://arnoldit.com/wordpress/2008/06/04/wikia-search-social-search-is-blooming/ by Stephen E. Arnold Wed, 23 May 2012 12:32:16 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: Summer Baby Monitor http://arnoldit.com/wordpress/2008/06/04/wikia-search-social-search-is-blooming/comment-page-1/#comment-80000 Summer Baby Monitor Tue, 17 Nov 2009 15:11:41 +0000 http://arnoldit.com/wordpress/?p=551#comment-80000 <strong>Summer Infant Extra Camera for Deluxe Day & Night Handheld Color Video Monitor...</strong> From the Manufacturer For parents wanting to monitor additional rooms or children in their home there’s the Summer Infant Extra Video Camera. Keep a watchful eye on your child wherever they are. The camera is wall-mountable, adjustable f... Summer Infant Extra Camera for Deluxe Day & Night Handheld Color Video Monitor…

From the Manufacturer
For parents wanting to monitor additional rooms or children in their home there’s the Summer Infant Extra Video Camera. Keep a watchful eye on your child wherever they are. The camera is wall-mountable, adjustable f…

]]> By: Stephen E. Arnold http://arnoldit.com/wordpress/2008/06/04/wikia-search-social-search-is-blooming/comment-page-1/#comment-13272 Stephen E. Arnold Sun, 08 Jun 2008 14:22:36 +0000 http://arnoldit.com/wordpress/?p=551#comment-13272 Thanks for the thoughtful post. The ideal social search may require total monitoring, analysis, and autonomous integration into the search system to be useful. Partial solutions--such as analyzing a chunk of traffic from GPS enabled phones--produce some useful, but overly general data. The tighter the monitoring of the social set, the more useful the data's granularity. Statistical projections based on samples or "what's generally available" can be useful in certain contexts. But my experience suggests this precept: "Monitor, capture, and analyze everything." That view scares this squawking duck, but certain countries and companies are taking baby steps in this direction. Stephen Arnold, June 7, 2008 Thanks for the thoughtful post. The ideal social search may require total monitoring, analysis, and autonomous integration into the search system to be useful. Partial solutions–such as analyzing a chunk of traffic from GPS enabled phones–produce some useful, but overly general data. The tighter the monitoring of the social set, the more useful the data’s granularity. Statistical projections based on samples or “what’s generally available” can be useful in certain contexts. But my experience suggests this precept: “Monitor, capture, and analyze everything.” That view scares this squawking duck, but certain countries and companies are taking baby steps in this direction.
Stephen Arnold, June 7, 2008

]]> By: Daniel Tunkelang http://arnoldit.com/wordpress/2008/06/04/wikia-search-social-search-is-blooming/comment-page-1/#comment-13153 Daniel Tunkelang Sat, 07 Jun 2008 20:19:03 +0000 http://arnoldit.com/wordpress/?p=551#comment-13153 Collaborative tagging / social search is a powerful idea to leverage the wisdom of crowds, but it yields mixed results. In particular, the resulting tags are generally sparse and inconsistent. For example, consider <a href="http://del.icio.us/tag/socialnetworking" rel="nofollow">http://del.icio.us/tag/socialnetworking</a> vs. <a href="http://del.icio.us/tag/socialnetworks" rel="nofollow">http://del.icio.us/tag/socialnetworks</a>. And this problem is even worse inside the enterprise, where there is typically a smaller user base than on the open web. In my experience, collaborative tagging is a great starting point for aggregating the wisdom of crowds, but by itself cannot produce index terms that achieve sufficient precision and recall to be useful for retrieval or exploration. The critical next step is to leverage this knowledge across the corpus. Regina Barzilay's group at MIT is doing interesting work in this area, as is Endeca. Collaborative tagging / social search is a powerful idea to leverage the wisdom of crowds, but it yields mixed results. In particular, the resulting tags are generally sparse and inconsistent. For example, consider http://del.icio.us/tag/socialnetworking vs. http://del.icio.us/tag/socialnetworks. And this problem is even worse inside the enterprise, where there is typically a smaller user base than on the open web.

In my experience, collaborative tagging is a great starting point for aggregating the wisdom of crowds, but by itself cannot produce index terms that achieve sufficient precision and recall to be useful for retrieval or exploration. The critical next step is to leverage this knowledge across the corpus. Regina Barzilay’s group at MIT is doing interesting work in this area, as is Endeca.

]]>