I've spent a good number of years on the Internet and one of my serious gripes has always been about search. From the early days of Yahoo search, there has been a discontinuous series of improvement of how search engines strive to be more accurate etc. Which meant that along the journey of Internet Evolution, I hopped from one search engine to another trying to find which one suits my needs the best. What a journey that was. Moving from Yahoo to Altavista, Dogpile, Teoma, Looksmart, Hot Bot, Vivisimo and even the graphical search results of Kartoo (Which is still around and still pretty awesome for an alternative way to search). Note that I have lumped all these names under the loose term 'Search Engine'. Some of them are meta-crawlers. That is a subject for another day. :)
Kartoo: Graphical search results ..mmmmm
Heck, I even wrote an entire chapter on how to optimize search, search engines and metacrawlers for research in our company training manual (for sales people and program developers). However as the years went by I started having some hope that search will be more accurate and I don't have to use Boolean operators, quotation marks and the like. That was partly satisfied as Google refined their algorithms and currently, I use their excellent, unbranded experimental search engine, searchmash.
Somewhere last year however, a project was dropped into our (as in my company) lap. Titled "Semantic Search" it was described as a more accurate, more user friendly and natural way to search. Unfortunately the project got dropped. (we're still trying to make time to design around Google's search algorithm patents tho!) So what is the meaning of this 'Semantic Search'? For that you gotta have a little understanding of how conventional search works. The most basic way an internet search works when a search request is sent is that an algorithm searches, computes and returns the highest number of occurrences of that particular search term. The rest of the programming in the search engine then displays the pages that have that number. (Disclaimer: that's how I understand it. I don't work in this industry and have no deeper knowledge of this unless I go do more research which i don't have time for. Dammit Jim! I'm a Sales guy not a Google employee!)
How is Semantic Search different then? Instead of searching, computing and returning the occurence of the search term, a Semantic Search algorithm actually looks at the meaning of the search term and then displays results relevant to the meaning. Still confused? Check out this video:
So I heard some mention of Semantic Search this morning on Buzz out Loud (ep 722) when the Awesome Threesome (Tom, Molly and Jason) discussed a website called Powerset. This was based on an article written by news.com writer Dan Farber on his 'Outside the Lines' column. Powerset is a search engine that uses Semantic Search algorithms to search. However it only searches Wikipedia and Freebase.com at the moment. Relatively new, their 'About Us' claims that their goal "is to change the way people interact with technology by enabling computers to understand our language". Also, "In the search box, you can express yourself in keywords, phrases, or simple questions. On the search results page, Powerset gives more accurate results, often answering questions directly, and aggregates information from across multiple articles."
So far so good. A video demo of what Powerset can do can be found here:
The official Powerset blog is here:
Ok, Ok. On to the hands on test. Here are some screen shots of some search terms I tried:
Overall, I was very impressed. It almost felt like I had a magnifying glass to Wikipedia. Again, the technology is very new and Powerset utilises technology licensed from PARC's Natural Language Processing patents . While this may not immediately shake Google's entrenched position, it will up the ante in terms of what needs to be incorporated into future search engines. An interesting observation that I recall from a few years ago is that most people will willingly embrace new search engines...for a few minutes and then return to the one that they've always been comfortable with. Whatever the case maybe, I hope this would be a good direction for search companies to take us into the future of search
Meanwhile, Jason Calacanis (who I think is cool!) has written his thoughts on Powerset (better written than mine). I find this blog post very interesting as he offers some good counter points and a more in-depth look at Powerset. You can read his views here.
Some more links for you to read:
http://www2003.org/cdrom/papers/refereed/p779/ess.html
http://en.wikipedia.org/wiki/Semantic_search
Wednesday, May 14, 2008
{ } Powerset - Search just got a little Smarter.
Labels:
Buzz Out Loud,
Google,
Jason Calacanis,
Powerset,
search engines
blog comments powered by Disqus
Subscribe to:
Post Comments (Atom)