by Mary Ellen Bates
The search engine Gigablast (http://www.gigablast.com) has been around since 2000, and generally doesn’t get a lot of attention. It’s a search engine for search-logic geeks — you can do some very subtle searches that aren’t supported in most other search engines.
Let’s start at Gigablast’s main page, which features a goofy-looking rocket ship. Click the Advanced Search button to take advantage of the usual search engine features — the ability to combine AND, OR and NOT (you must put the Boolean operators in ALL CAPs); phrase searches, site search (“restrict to this URL”), and link search (“pages that link to this URL”). You can also specify for each search how many results you want per search-results page (10, 20, 30, 40, 50 or 100. (You can also use a pull-down menu on the search results page to change the number of results per page — a nice
You can even “restrict to these sites” — limit the search to just the sites you list on the Advanced Search page. It’s similar to the site search but with the ability to specify multiple sites. So, for example, if you were searching on a health-related topic and wanted to just look at the sites you consider most authoritative — say, mayoclinic.com, nih.gov, health.yahoo.com, and webmd.com — you could restrict the search to just those sites.
You will notice that there is also an option on the Advanced Search page to select “site clustering”, which does what most search engines do — includes one page from a site on the search-results page and indents the other pages from that site. Try turning “site clustering”
off, and you will be able to see ALL the pages retrieved from a site, not just the few that appear on the search- results page.
One of the nice features in the search results page is a list along the top containing “Giga Bits” – the most common words or phrases that appear in the top search results, with links so you can AND that word/phrase to your search to narrow it down.
But the geekiest search features of Gigablast aren’t even evident from the Advanced Search page; you have to read the Help files to find out about them.
Here is a list of the features I consider the most cool:
* If you type two words in the search box (seattle books, for example), the default of Gigablast and most other search engines is to give preference to sites that have those two words next to each other. However, in Gigablast, you can “disable” that preference by typing seattle .. books It’s an AND operator but with no higher ranking for sites that have “seattle books” as a phrase.
* If you type (nanotechnology | manufacturing) Gigablast looks for sites that have nanotechnology and manufacturing in the page, but the relevance ranking is solely on the word manufacturing. In other words, it first executes a search for nanotechnology, then does a search within the results for manufacturing and ranks the results solely on manufacturing.
* Gigablast has a “squishy Boolean” logic feature, meaning that you can specify which words the web page must have and which words would just be nice to also have. So, for example, the search vaccine OR avian AND flu is interpreted by Gigablast as saying “All search results mush have both avian and flu in them. Rank higher the results that also have the word vaccine in them.” (Or for the logic geeks like me, it is interpreted as [avian AND flu] AND [vaccine OR avian].)
One other interesting service through Gigablast is called Custom Topic Search (http://www.gigablast.com/cts.html).
This lets you create your own subject-specific search engine, limiting the search to up to 500 of your favorite URLs. You can then insert your specialized search engine on your own web page and visitors to your page can search just in the sites you consider most authoritative. You can see an example of the Custom Topic Search at GigaUFO, which limits your search to the most reliable UFO-related sites. (http://www.larryhatch.net/GIGAUFO.html)
A version of this Tip with live links is available at www.batesinfo.com/tip.html An RSS feed for my Search Tip of the Month is at www.batesinfo.com/tip.rss