Powerset Nails Search: A Very Bold Assertion
June 29, 2008
Chris Gaylord, writing for the Christian Science Monitor, updates a May 2008 essay, and emphasizes this point:
Google has been a bit dismissive of semantic search, preferring (for now at least) its quick keyword approach. But this Microsoft news puts a lot of weight – and $100 million – behind the notion that web users want to ask questions to a search engine, not just feed it keyword clues. We have yet to see if Microsoft will keep the Powerset name or, more likely, integrate the technology into its Live Search. That site certainly needs some help. The company has fought a losing battle against Google and Yahoo for years now. Despite its best efforts and even cash incentives, Microsoft has not been able to distinguish itself. Offering a strong semantic search option is a good way to reboot the challenge.
You can read the full document here.
You may recall that the original Ask Jeeves answered questions. Human figured out answers, put them in a file, and the Ask Jeeves’ system converted the user’s query to a form that could be matched against the canned answers. The buzz about this surged in the late 1990s, but the cost of the Ask Jeeves’ approach was high, and in my view, the system did not work very well.
The desire of information retrieval mavens to take a question, any question, and have software answer it makes some folks darned excited. The technology to answer questions continues to advance, and it is possible to get answers from a number of different systems. I have participated in meetings where smart people much more enthusiastic than I argued about the importance of having a system answer a question.
I have written about NLP or natural language processing in the first three editions of the Enterprise Search Report, and I added some information in my April 2008 Beyond Search study for Gilbane Group. Let me offer some observations:
- I don’t type queries into search engines. I prefer Boolean statements and point-and-click interfaces that let me “see” what’s in an indexed corpus. My experience is that typing questions is not too popular, nor is the notion of chopping text from an article and letting a search system find “more like this”. I have an installation of the Brainware trigram system, and it is useful–far more useful to me than asking “When did Columbus discover America?” if indeed he did. No NLP system can make much sense of a short query in the context of archaeological research about pre-Kit Columbus visitors to the North American landmass. Nope, that type of question answering will take a bit more lab work.
- NLP imposes considerable computational load on both the document indexing subsystem and the query processing subsystem. I saw an impressive set of PowerPoint slides at the 2007 BearStearns’ Internet conference, and I fiddled with the Wikipedia demonstration in 2008. What I have not seen is proof that Powerset’s amalgam of Xerox technology and its proprietary code scaling. Without scaling, NLP is likely to remain interesting but of little use to me.
- Microsoft, like Yahoo, is now in the business of collecting search technologies. There are two “flavors” of SharePoint search. There is the Fast Search & Transfer technology. There is the whizzy new Live.com search. There is search in XP, in Vista, in SQL Server, and probably other search technologies I don’t know about. Toss in Powerset. What the collection resembles is a yard sale, not an exhibit of Etruscan tomb art at the British Museum. Search has to be more than a yard sale in its design, architecture, and technical framework. The cost of integrating this stuff is more than my check book can support.
I appreciate the enthusiasm for Microsoft becoming more competitive. Let us not forget that Google has been doing pretty much the same thing–it’s one trick pony show–for a decade. With two thirds of the market for Web search, Microsoft has some work to do to become a number two in search. Google continues to seep into the enterprise via osmosis. Let’s face facts. Customers have to buy from Google. Google is not very good at sales, customer support, or communicating what its gizmos can do. Microsoft is a good sales organization, but it is watching Google challenging its enterprise revenue the way spilled ink spreads on a white table cloth. And, Google has serious semantic technology which is a widget in a larger data management solution at Google.
Keep cheerleading for Microsoft. Just keep the challenges of NLP in mind. Agree? Disagree? Let me know so I can learn what I don’t now know.
Stephen Arnold, June 30, 2008
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
