September 2003 Archive

Personalization in Google's Future?

Today, Google announced its acquisition of Kaltix Corporation. Formed just this past summer in June, Kaltix has been working on developing search technologies related to personalization and context-sensitive searching. What Google will actually do with this technology remains to be seen and will likely take awhile before it is implemented for the public. And given Google's number of products, if the technology is used, it might be for their ads, news, or their shopping database rather than for their general Web search engine.

Invisible Web Directory Disappears

I just noticed that Intelliseek's InvisibleWeb.com directory of hidden or invisible Web databases and resources now just redirects to Intelliseek's ProFusion meta search engine. It looks like they may have made the change on Aug. 13 or earlier. While the directory had not been updated in awhile, it was a useful as place to look for some hidden Web resources, especially databases. Some of those databases are still searchable within ProFusion's search groups, but it was the directory function that was most valuable. For a replacement, try Invisible-Web.net from the book by Gary Price and Chris Sherman or CompletePlanet instead.

New Yahoo! Shopping Search & Tab

Following yesterday's Shopping.com relaunch, Yahoo! has today relaunched their shopping search. It is available both at shopping.yahoo.com and as a new tab from regular Yahoo! searches. The new tab has a "Products" label and is on the far right just after the News tab. Like Shopping.com, results include product information, user reviews, and a price comparison feature. Yahoo! shopping even has an alert feature, to send email alerts when prices change.

Yahoo! has fewer narrowing and sorting options, but its underlying database draws primarily upon the Yahoo! stores (plus outside merchants who pay to supply product feeds to Yahoo!). This is a different collection of merchants than Shopping.com. Sometimes Yahoo! has more stores listed, and sometimes Shopping.com has more. So who has more and who has the best price. On a quick search comparison for two digital cameras, it varied. For the Olympus C-5050, Shopping.com had 68 stores compared to Yahoo!'s 75. Yahoo!'s search had the best price by $1.05. For the Olympus D-390, Yahoo! had only 34 stores to Shopping.com's 42. This time, Shopping.com had the better price by $14. None of these price comparisons include shipping or other extra charges.

And how do both compare to Google's Froogle? Froogle finds many more hits, but they are not all necessarily hits from the actual product. For these cameras, they are often for accessories such as batteries or memory cards. And in both cases, the least expensive actual camera that Froogle found was more expensive, than the cheapest at both Yahoo! Shopping and Shopping.com.

Read more about the Yahoo! changes in the press release.

Dealtime + Epinions = Shopping.com

Shopping.com has launched anew as a combination of the old Dealtime and Epinions sites. This newly relaunched shopping search engine includes the price comparison features and product meta data from Dealtime and consumer reviews from Epinions. Both of the other two sites will continue to operate separately. The shopping.com search gives narrowing options for price, brand, and product-specific features. It also offers several sorting options such as price, rating, relevance, and store rating. These narrowing and sorting options are a welcome change from the most general Web search engines that offer no such sorting capabilities. The site is most useful for commonly-available products since it has a fair number of larger stores in its database. I found zero hits on some more unusual products. The database scope is limited in that there are many online stores not included. Read more in their press release.

GeoSearch at Google

Google is new experimenting with a new “Search by Location” in Google Labs. They are finally catching up with a feature that the old Northern Light had years ago. Google has added a map of locations for the hits from MapQuest. It highlights matching addresses in the keyword in context (KWIC) display, but there is no cache link. At this point, it seems to be limited to U.S. addresses. Searches must include some address information. Full state names and ZIP codes appear to be normalized to a city, state abbreviation search. In other words, the address can be entered in a variety of formats, but the KWIC highlighting usually only highlights the city and state abbreviation in the record, at least on the searches I tried.

This is still very much an experiment. It may not always be available. It may change greatly. Certainly at this point, some of its matches for locations are quite inaccurate.

Geo Meta Tags at Gigablast

Meta tags, originally introduced to the search space by AltaVista as way to get better descriptions of a Web page, were terribly abused by search engine spammers. Only the meta keywords and meta description tags were indexed, and only by some search engines. Now, Gigablast is introducing new meta tags and is daring to try again. This time, they are "geotags," meta tags for identifying the location of a Web site. According to the information page, the following meta tags are now indexed and supported at Gigablast:

• zipcode
• city
• state
• country
• author
• language
• classification

Before this will become very useful, Web sites will need to start using the meta tags. Note that this does not use the Dublin Core meta tags. The same post also announces a faster refresh cycle and a doubling of the Gigablast database: "I am planning a significantly faster update cycle, not to mention growing the index to about 400 million pages, all hopefully in the next few months."

Two More Google Experiments: Related Searches and Spectrum

The ever-experimenting Google has added two more experiments that a very small portion of their users may see. First, they are finally experimenting with giving suggestions for "related searches." This is one feature that they could have added long ago and that many other search engines have offered for years. But the few early experiment reports have not been very impressive and seem to have poorly related suggestions. Presumably this will be much improved before it is released, if they ever release it at all.

Then there is Google's Spectrum, which is a Google search counter. Users can see how many Google searches they run each day. It not yet publicly available, but according to Google is just "an experiment we've just started running with a small sample of Google users."

Both of these are not available to the vast majority of Google users at this time.

Update on Gigablast Boolean

The full Boolean capabilities of Gigablast announced on Monday don't always seem to work right. The - is working more accurately than either NOR or AND NOT today. I am hoping it is a momentary glitch since I just updated the Gigablast review, the search feature chart, and the search engines by search feature page.

Search the Wayback Machine with Recall

The Internet Archive has released a full text search capability via its new offering called "Recall." The basic interface is available at web.archive.org while the advanced search is at recall.archive.org. The advanced search adds date limits.

This is a beta offering, so it may not always work. It claims to search a huge portion of the Internet Archive's Wayback Machine database covering about 11 billion pages. Note that the Wayback Machine offers browsing by URL access to 30 billion pages.

Boolean at Gigablast

It may be a small search engine, but Gigablast keeps on innovating. Matt Wells announced today the support for Boolean operators at Gigablast: AND, OR, AND NOT, and OR NOT. It is also supposed to support nesting. It is available from the main page search box. Operators should be in all upper case.