June 04, 2008, 8:45 PM — As startups and established players strive to develop Google-killer search
technology, conceptual differences in their approaches make for interesting
discussion.
Case in point: Jimmy Wales' Wikia
Search and rival Powerset,
which is using Wikipedia,
the free online encyclopedia project Wales founded, to argue that its semantic
engine represents the future of search.
During an interview this week, Wales was asked for his opinion about Powerset
and he declared himself unimpressed, at least so far.
For starters, finding content on Wikipedia is quite easy, he said. "I
find that search at Wikipedia works perfectly fine." Plus, indexing Wikipedia
content doesn't pose a major challenge, Wales said.
When Powerset unveiled a test version of its much-awaited semantic search engine
last month, Wikipedia played a major role in the marketing push as one of only
two Web sites included in the index.
While acknowledging that the scope of its index was extremely limited, Powerset
executives said that the engine's ability to -- in their view -- improve Wikipedia
search reflects what it will do later for Web search in general.
Wales isn't convinced.
"It's really hard to judge right now [the quality of Powerset's engine]
because searching Wikipedia is a pretty easy thing to do. It doesn't present
much of a challenge. Wikipedia isn't a very large data set and it's a pretty
simple thing to do, to index Wikipedia," Wales said. "So whether their
approach is going to be useful on a bigger data set [is hard to tell]."
It will be interesting to see how Powerset's technology evolves, said Wales,
but he added that he isn't sold on semantic search technology in general. "I
haven't been very persuaded so far by what I've seen about the semantic approach.
At least so far, I'm not that interested in it," he said.
According to Powerset, its users do find value in its Wikipedia search. Because
Powerset can "understand" the pages it indexes, it can do more than
return the proverbial 10 blue links for search results. For example, it can
assemble a collection of facts related to the query, as well as summarize the
found information. It can also provide direct answers to factual questions.
"Our early users tell us that Powersets automatic extraction and
aggregation of key facts about topics is extremely useful, since that information
is often strewn across many different pages in Wikipedia," said Scott Prevost,
Powersets general manager and director of product, via e-mail.
"We've also received a lot of positive feedback about our automatically
derived summaries of each Wikipedia page, which helps users to scan a pages
content and easily navigate to relevant parts of the text," he added.
Prevost also defended the promise of semantic search, which is designed to
extract meaning from the Web pages it crawls, as opposed to focusing on keywords,
which is the approach of all major search engines, including Google's.
"Semantic search is in its early stages, and is already showing great
promise. The real power of semantic search is not only in providing search results
that better align with the meaning of the search query, but also by providing
new ways for users to explore and navigate information," Prevost said.
One thing that Wales and Prevost share is the belief that current Web search
technology can, and should be, significantly improved.
Wales this week unveiled the latest version of Wikia Search, a search engine
developed by his company, Wikia Inc., that hopes to rely primarily on volunteers
for building and maintaining the index and determining results' ranking and
relevance, in a similar way as Wikipedia did in the online encyclopedia space.
Whether Wikia Search, Powerset and other new entrants to the search engine
market will be able to mount a credible challenge to Google remains to be seen,
but it will certainly be interesting to see them try -- and, in the meantime,
argue in favor of their respective approaches.














