inteligentne (?) wyszukiwarki kontekstowe

[pneuma]

Czy używacie takich wyszukiwarek, czy w nich pozycjonujecie?

jakie to?

- polska próbka:

https://carrot.cs.put.poznan.pl/carrot2-web...oller/index.jsp

- dosyć znana zagraniczna: :roll:

https://www.eurekster.com

jaka przyszłość dla nich?

[melassa]

carrot wydaje mi sie ze tylko zassysa wyniki googla jak narazie

ta druga nie mam pojecia ale jesli wszuskuje na podstawie kontekstu poprzednich przeszukan to dosc ciekawe rozwiazanie takie proby juz robi google i ms

[pneuma]

Ukazał sie bardzo ciekawy artykuł dotyczący wyszukiwania kontekstowego, które ma stać się "zabójcą takich wyszukiwarek jak google.

Google killers 

wtorek 20, lipiec 2004 

By Simon Sharwood

Can Google’s armoury fight off the swelling ranks of rivals? Simon Sharwood checks out the future of Internet searching.

Can new weapons in Google’s armoury fight off the swelling ranks of rivals? 

This article is featured in APC June 2004 Back issues

SubscribeIs it possible to out-Google Google? While the search engine’s patented PageRank system is legendary among Web surfers, some say new search technologies on the horizon will blow away current search concepts.

A vicious search technology arms race is under way which is giving rise to a crop of new contenders fighting over Google’s crown. Microsoft has Google in its sights, and startups such as Mooter believe “nodal maps” are the way forward. Through algorithms, page ranking formulas and sophisticated new search options, these new rivals are attempting to create a brave new post-Google vision.

“During Google’s honeymoon phase, it exceeded users’ expectations,” argues John Brand, vice-president of technology research services at analyst firm Meta Group. “But as people got used to the Google experience, they began to want more.” 

For many people, that “more” is more speed — results even faster than the half-second or so Google typically takes. Others want better coverage of dynamic sites where content is stored in databases and hidden from spiders and bots. This would take the number of pages covered by a search engine well beyond the 

6 billion or so Google claims to cover today.

“Traditional linear engines are hard work for our minds,” says Liesl Capper, CEO of Australian search engine Mooter.com. “We read the results and then go back and re-educate the search engine with a better query based on what we have learned it will return.” 

Capper describes this process as frustrating, time-consuming and counterintuitive. “Humans are hard-wired to form conceptual groupings for information,” she says, adding that most search engines simply don’t cater to this deeply ingrained need. 

The new Google

The search race is inspiring new ways of thinking among Google’s developers, including two new search modes called “personalisation” and “social networking”.

The personalised search is important because some searches are intrinsically difficult. A search for “apple”, for example, could be a search for a computer or for a fruit.

Personalised search services collect information about the subjects which interest you, so a search engine knows in advance that you are interested in computers and not fruit. Google’s version of personalisation does just this. After completing a profile, searches made at https://labs.google.com/personalized produce not just the usual list of results but also a “personalisation slider” to apply your profile to search results. 

Set the slider to maximum and all results are subjected to an analysis based on your profile. Select minimum and the results revert to Google’s standard service. Set the slider to somewhere between these two extremes and it will adjust the balance to reflect the placement.

The flipside for Google is that your profile also has the potential to allow it to charge more for advertising. At present, the profile is stored on users’ hard drives as a cookie and advertisers’ greedy mitts are kept away from its potential for targeted advertising.

The company’s other new technology is a social networking service called orkut.com (www.orkut.com). The site aims to “expand the circumference of your social circle”. Google is currently tight-lipped about orkut.com and its technology, stating only that it was created by employee Orkut Buyukkokten under its policy of directing engineers to spend 20% of their time on side projects of possible benefit 

to Google users.

Testimonials on the Web report that the sign up process for orkut.com collects a lot of personal data and offers services very closely modelled on existing social networking services, but adds stronger user-authentication tools so customers can feel confident their new online buddies are real people.

The site’s FAQ also hints at its commercial possibilities, stating: “We will never rent, sell or share your personal information for marketing purposes, except with your express permission. We may however share your information with our affiliate, Google, in order to analyse how our members use the site and how we can improve orkut.com or develop new features.”

Social networking’s current pin-ups are meetup.com (www.meetup.com) and friendster.com (www.friendster.com). Both sites enable strangers to meet up online with profiles of their interests and activities acting as bait for like-minded people to strike up a relationship. Both also use these profiles to create targeted advertising.

“Local search” is another important innovation. Currently offered in the US only, this narrows searches by geographical location to produce even more relevant results and make Google a potential rival to Yellow Pages businesses worldwide.

The Competition

Faced with Google’s multiple escalations, Yahoo has responded by developing and building its own technology.

For competitive reasons, Yahoo is cagey about the exact technology involved. However, Peter Crowe, the company’s Australia and New Zealand business manager for search, said the underlying technology relies on “ranking the relevance of a site in relation to lots of other sites. If a lot of sites link to you, that is an indication that it is an important site relative to another.” 

The company operates a “content acquisition program” to trawl dynamic sites, usually at the behest of companies willing to pay for the service.

Yahoo is also investigating social networking. Having already made it possible for people to share searches through its popular instant messenger service, it now plans to examine other ways for its members to interact based on their common interests.

Microsoft has repeatedly stated it will introduce its own search engine this year. A crawler-based search technology project called MSNBot seems to be at the heart of its efforts and the company has stated it is investigating personalisation services. 

Natural language search — technology which allows searches to be phrased as questions such as “Who wrote War and Peace?” — is rumoured to be in Microsoft’s sights, as is tight integration with an enhanced version of Internet Explorer and eventually with the next version of Windows.

Redmond watchers are tracking a site erected by Microsoft Research Online Lab called Wallop (www.mywallop.com). Touted as a place where “you can share photos, blog and interact with your friends”, the site seems to offer social networking services and is said to integrate with Microsoft Instant Messenger.

One service that is unequivocally about search is Mooter, focusing on new types of search interfaces. Mooter does not present results as a list, instead offering a “nodal map” which groups results by themes. 

For instance, a search for “Australian Personal Computer magazine” produces a hub-and-spoke map with the phrases “Internet”, “apc” and “personal” returned as three of seven initial nodes. Clicking on the node called “apc” then produces a conventional site list.

Mooter uses a method that Liesl Capper calls “implicit search” to “find the underlying pattern behind the search keyword instead of just returning the literal meaning of the words or phrase.” While the site’s traffic is light in comparison to Google, Capper says that Mooter’s users seem to be “a smart and powerful niche,” an assertion supported by the fact the site’s top 10 search list often includes phrases such as “gas centrifuge” instead of pop-culture terms like “Janet Jackson’s breast”.

John Brand at Meta Group classifies Mooter as an “ontology engine” and says that such sites may well find substantial audiences once the technology matures. But Markus von der Leuhe, a senior analyst with Internet ratings company Red Sheriff, believes that Google, Yahoo and Microsoft are 

now unassailable as the three giants of search, and that any changes in the pecking order will come as these new contenders and others scrap for places four through to 10.

Yet even the three search giants may be barking up the wrong tree. “The theory that personalisation will attract a large audience has not been proven,” says Brand. He is also sceptical about social networking because he believes the investment in analytics tools required to create targeted advertising exceeds potential revenues.

Brand highlights the fact that while this second search arms race promises to offer better search results, the technologies are designed give search engines a better story to tell to advertisers.

History shows that no matter how popular the dominant search, it’s always possible for the leader to lose its edge. For a few months in 1995 and 1996, the hottest search engine was Alta Vista, a company which joined the long list of dot com casualties when the search race erased its technological edge.

This time the objective seems to be revenue, not results. Time will tell whether Google users will be better or worse off for it. 

Search Rivals

Grokker

How it works: uses a US$49 proprietary client to present search results as maps, displaying content in a separate window.

Useful for: browsing lots of links without hitting the back and forward button dozens of times.

Web link: www.groxis.com

Vivisimo

How it works: conventional index but presents results clustered into topics listed in a directory tree.

Useful for: refining a search without ever having to read the results.

Web link: www.vivisimo.com

Teoma

How it works: by using technology called “subject-specific popularity”. This determines the best answer for a search by asking experts within a specific subject community about who or what they believe is the best resource for that subject.

Useful for: a left-field second opinion.

Web link: www.teoma.com

Eurekster

How it works: invite your friends to join the engine, which then matches the results you all appreciate to refine results.

Useful for: getting your friends’ perspectives on relevant results.

Web link: www.eurekster.com

Mooter

How it works: presents results in nodal categories designed to help your brain decide which links are relevant.

Useful for: getting a fresh view on a search.

Web link: www.mooter.com

A9

How it works: takes Google’s results, presents them in a different interface, adds the ability to search in Amazon.com’s vast library of digitised books and tracks all past searches for the entire database of Amazon.com customers.

Useful for: searching through books as well as the Web.

Web link: www.a9.com

źródło