Pandia Search Engine Weblog September 2003

(September 30 2003) In August Pandia reported that a spin-off company from Stanford was developing a new technology that could lead to the birth of a new search engine.

The idea behind the company was to personalize the Pagerank system, so that the search engine could learn from searcher behavior.

Kaltix was formed by three members of Stanford's PageRank team, the research environment that gave birth to Google. This is why it now comes as no surprise that the company has been acquired by -- you guessed it -- Google.

"Kaltix is working on a number of compelling search technologies, and Google is the ideal vehicle for the continued development of these advancements," says Larry Page, co-founder and president of products at Google.

Press release.
Webmaster World discussion.
An Analytical Comparison of Approaches to Personalizing PageRank (a science paper written by members of the Kaltix team, PDF-file)

Amazon to launch shopping search engine

(September 28 2003) This week Yahoo! officially launched its new shopping search engine. Yahoo! lets you search a large database of products from merchants affiliated with the company.

As previously reported by Pandia Google is still working on its Froogle product search search engine. European Kelkoo is testing a similar service, and the American old-timer Dealtime has launched the new Shopping.com comparison shopping search service (see below).

Online shopping is clearly seen as an important revenue generator.

No wonder then that Amazon.com would like a larger piece of the pie. What was originally a bookstore has expanded into all kinds of goods, from CDs to kitchen appliances.

It has also made deals with a lot of other online vendors in order to be able to offer as many products as possible.

Amazon will now also launch its own shopping search engine, going under the nickname of "A9". The plan is to launch the new site on October 30, and Amazon has already some 30 people working on the project.

It is unclear whether the name of the new search engine actually will be "A9". The A9.com domain is registered to one Wayward Jennings , not Amazon.com, and resolves to an error page.

Canadian Press has more.

Verisign takes control

(September 25 2003) Some of our readers have asked us what the new default "Site Finder" search engine is. It appears every time someone enters a non-existing domain name in the browser address field.

The Sitefinder search engine is the result of a controversial move made by Verisign, the company that administers the .com and .net domains. It hijacks all request that goes to non-existing domain, and delivers its own default "error page" to anyone who misspells a URL.

Google and AltaVista test local search

(September 23 2003) Q: What's the next big thing in internet searching? A: Localized search.

The search engines are now testing technologies that let you narrow your search to one local geographic area, being that a city, a county or a state.

Read Pandia's article on geo-targeted search engines!

Gigablast test geographic metatags

(September 22 2003) While most search engines considers web page metatags (i.e. special tags signifying the content of the page) for dead, newcomer Gigablast is experimenting on new so-called "geo-sensitive metatags".

The idea is that web site owner shall help the search engine target shops and services in one particular city of geographic area. Gigablast gives the following examples:

There has been made many attempts using meta tags as the foundation for web page classification and sorting. The problem has always been that unscrupulous webmasters have used the metatags to spam the search engine indexes, mostly by including false information. Due to this any innocent search query could ultimately bring the searcher to an "adult" destination.

If Matt Wells, the owner of Gigablast, wants this to work, he will ultimately have to add algorithms that test the validity of the content of the new meta tags. Then, of course, he will have to convince webmasters all over the world to use them.

By the way, Resourceshelf reports that Wells is planning to improve the Gigablast update cycle and double the size of the index in the next few months.

Here's a bet: How long will it take before one of the large search engine companies buy Wells' technology? Two months? Six months? A year? We'll keep you posted.

Dealtime becomes Shopping.com

(September 22 2003) The American comparison shopping company Dealtime has changed its name to Shopping.com. The new site will integrate Dealtime's directory with product reviews from Epinions, a site Dealtime acquired in May this year.

Like all comparison shopping search engines Shopping.com lets you compare prices from various merchants. However, you may also click on a link to an Epinions product review, thus gaining more information before you buy. The reviews are written by Epinions visitors.

The old Dealtime and Epinions sites live on.

We have not had the time to test the new Shopping.com site yet. The color scheme definitely proves that the 70's design revival has reached the US, though. We don't know how we feel about that.

Chris Sherman of Search Engine Watch has more.

FindWhat's cold feet

(September 19 2003) As reported by Pandia this summer the American pay per click text ad search engine FindWhat announced that it would buy European Espotting, thus laying the foundation for what could become a new major player in this market.

Alas, there seem to be a strain in the relationship. FindWhat now wants to renegotiate the deal. After having reviewed Espotting financials FindWhat has changed its views regarding Espotting's strengths and prospects, and could even decide not to finalize the merger.

Pandia does not know why FindWhat may want to leave its European bride at the altar, but one reason may be the hard competition in the European market.

Given that both Google and Overture now can offer complete packages combining regular search results with pay per click text ads, Espotting might find it harder to compete. Espotting does not own a regular search engine of its own, nor does the company have a reliable search engine partner.

FindWhat.com says that, at a minimum, any successful renegotiation of the terms will include "a reduction of the purchase consideration, along with other material changes to the existing Merger Agreement."

See also: Reuters and FindWhat press release.

Text ads bring in the cash

(September 18 2003) It is ironic that in the age of high tech wizardry it was the plain text ad that ultimately saved the search industry, text ads disguised as search results that is.

Pandia takes a look at the growth in pay per click search engine advertising.

Wotbot becomes Wotbox

(September 17 2003) As Pandia reported in August there is a new British search engine in town called Wotbot.

Actually, it is no longer called Wotbot. Lycos has threatened legal action, claiming that the name Wotbot infringes on their trademark rights to Hotbot. And indeed, the names are probably a little bit too similar.

"With repeated pressure and threats of legal action, we had very little choice," Mike Nott of Wotbot says. "We are a very small outfit - two guys, both with responsibilities, so the risk of losing our homes was too great to take."

Which is why the search engine has got a new name: Wotbox.. The old URL is still working.

Did you mean: Google?

(September 15 2003) Google is testing a new feature on some selected searches. The search engine adds some links, suggesting alternative and more focused search queries. The feature is called "related links" and is added to the top of the list of results.

According to GoogleGuy, a Google representative, an internal group came up with the idea for this experiment. That being said, quite a few search engines and metasearch engines have had similar features already, including Teoma, Lycos, AltaVista and AlltheWeb.

Google has not yet decided whether they will implement the new feature across the board. If it works, it can actually be a good idea, as it helps non-savvy searchers target their searches more efficiently.

By the way: Google is also testing a new format for their AdWords text ads, where the present separate colored boxes are replaced by a right hand column for continuos ads. Search Engine Lowdown has a screen shot of this alternative format. Webmaster World discussion

Kelkoo, Yahoo! and Google are making preparations for the shopping war

(September 8 2003) If we are to believe some experts, shopping will be the arena for the next search engine war. That is probably a slight exaggeration.

Nevertheless, surfers seems to be loosing some of their inhibitions as regards online shopping (giving away credit card information and all that), so a well established shopping search engine may bring in quite a lot of money through advertising and paid listings.

Yahoo! is testing out a new shopping search engine at products.yahoo.com. It is different from the old Yahoo! shopping section more or less in the same way as a directory edited by humans is different from a search engine that gathers results automatically.

The dividing line is a bit blurred though, as shopping search engines often rely on automatic XML or spreadsheet feeds from merchants, as do comparison shopping directories. However, as Yahoo! puts it "The Yahoo! Product Search engine also crawls the web pages of other, non-affiliated merchants to include their relevant products..." In any case, the way it presents results makes it look more like a search engine.

Google's Froogle shopping search engine also offer a mix of product information submitted by merchants and information gathered by Google's search engine spiders.

Google has recently made some small changes to Froogle, which is still in its beta test phase. They have added a left hand column giving the user access to forms and links that lets him or her sort listings according to price and gain access to the relevant directory category.

European Kelkoo has now presented a new revised version of its new search engine. This is actually also a test version, but Kelkoo is nevertheless using the new search engine by default at its UK site.

The entry page has become a little more complex than in the previous version, as it now includes a directory structure and a few selected products.

The search function is just as easy as in Froogle! and Yahoo!, however. Enter your product and Kelkoo makes a good guess at what you are looking for. The search result pages includes a pull down menu that let you narrow your search to a relevant category.

The winner of this race will be the company that offers an easy to understand interface and highly relevant results. And so far both Kelkoo, Yahoo! and Froogle is counting on simplicity.

Geo-targeted pay-per-click text ads

(September 8 2003) Speaking of shopping: Pay-per-click search engine giant Overture is now testing a new local search product allowing advertisers to "geo-target" their ads.

This makes it possible for searchers to look for a particular kind of shop or service in their local neighborhood. This is a very good idea indeed, as it expands the service from online shopping to regular advertising for "physical" shops and services. This means that your local cinema might find Overture a good alternative.

Searches for products in Norway bring us a map of Kansas. There are a lot of Norwegian-Americans there, they say. Never mind, Overture is not claiming that this is remotely like the final product, and US searches do bring results.

By the way, localization will probably be the next big thing in search engines. You heard it here first!

Webmaster World discussion
CBS Market Watch article

Google's five year anniversary

(September 8 2003) Congratulations Google! It is now five years since Larry Page and Sergey Brin founded their company. The rest is, as they say, history.

By the way, Pandia was born later that same year (1998).

Chris Sherman has more on the Google birthday. See also The Washington Post.

Gigablast adds Boolean searching

(September 6 2003) There is in fact a search engine world beyond Google, AltaVista and the other well known brands. Gigablast is one of them.

Gigablast is run by Matt Wells, an American mathematician. Wells originally worked with the search team at the now defunct Infoseek search engine before starting coding Gigablast.

A test version was ready in March last year, and the search engine has managed to spider some 200 million web pages.

Wells has now added true Boolean searching to Gigablast's advanced search functions, something which is bound to please librarians and other professional researchers. By searching for cats AND dogs for instance, you will get pages that include both words, not only one of them.

Boolean is a simple set of search terms that lets you combine keywords in a way the search engines can understand. The readers of Pandia can easily learn how to "speak Boolean" by reading the Goalgetter search tutorial. .

Gigablast asks you use to use AND NOT to exclude a word (cats AND NOT dogs). Alternatively you may - of course - use the minus sign instead (cats -dogs).

Please note that Gigablast defaults to OR, meaning that the search dogs cats will bring you pages that includes one or both of the words.

You do not have to learn Boolean to make use of Gigablast's advanced features, however. You may also make use of the site's advanced search form, which lets you combine keywords by using different form fields.

You may restrict searches to sites, URLs, titles, IP numbers, and links, as well as to PDF, doc, xls, ppt, ps and text files

Gigablast help pages
Search Engine Showdown on Gigablast
Source: Internetbrus and Resourceshelf

Search the past with Recall

(September 4 2003) Several historians have worried about the Net and the fact that web pages disappear just as easily as they are born. A lot of valuable historical material may be lost when a site shuts down or pages are deleted.

The Internet Archive is doing something about it. With regular intervals it sends robots our crawling the web, gathering data. It then keeps these "snapshots" of cyberspace in its computers, giving you the possibility to reconstruct the past.

If you search for www.pandia.com in the WayBack machine you will, for instance, get a list of snapshots of Pandia from May 1999 onwards.

Resourceshelf reports that The Internet Archive is now testing a new search engine that lets you search the contents of the 11 billion pages found in the index.

The Recall search engine is simple to use. Just enter your search query in the search form and hit enter.

Advanced search functions are included on all result pages. They let you limit your search to a specific time period. You may also narrow results by category and topics selected by Recall on the basis of the the meaning of the search query.

Please note: The links on archived search engine news pages will not be updated!

Search Engine News Weblog Archive September 2003

Google acquires Kaltix