Searching the Web
Volume 31, Number 1, October 2003
Clustering Search
Engines Holly Gunn
Clusters are groups that share common characteristics.
Clustering search engines arrange the search results
into groups around a common theme. Unlike most search
engines, which present search results in a linear list
based on relevance to the search query, clustering search
engines group results according to different areas of
meaning.
When searching the Web, information seekers
generally want search results that are highly relevant
to their
search query. However, there are occasions when clustered
results can be more useful than a list of relevant results.
They are particularly helpful when students are exploring
a topic for the first time and need to learn about various
subtopics, or when students are trying to zero in on
a specific aspect of a topic for further research. For
example, a search for fur trade using WiseNut (www.wisenut.com)
provides results arranged under headings such as History
Fur Trade, Fur Trade Forts, Hudson’s Bay Company,
Fur Trade on American Indians, etc.
Clustering search engines are useful when separating
results for ambiguous words or words with double meanings
or different uses, such as virus, spider, worm, dolphins,
shrimp, etc. For example, a search for dolphins with
Vivisimo (www.vivisimo.com)
separates search results about the Miami Dolphins from
those about the animal.
Clustering search engines: Meta-search
engines that cluster results:
- Vivisimo
(http://vivisimo.com)
Vivisimo, developed at
Carnegie Mellon, has received top ratings by Yahoo!
Internet
Life
in 2001and Search Engine Watch (http://www.searchenginewatch.com)
in 2001 and 2002. It is a good starting place for
students who are trying to decide on an aspect of
a topic for
further study.
- Guidebeam (http://www.guidebeam.com)
- iBoogie
(http://www.iboogie.com)
(This meta-search relies on advertising on its front
page and on pages containing search results.)
- Killerinfo
(http://www.killerinfo.com/m)
Killerinfo can search subject-specific channels,
health, business, reference, sports, science, government,
the
arts, government, news, and kids, as well as the
Web. Results are clustered on the right.
- QueryServer (http://www.queryserver.com)
This meta-search engine will send queries to the
News, Health, Money, and Government sites, as well
as to
the Web.
Many of these engines have “about” sections
at the bottom of their home pages explaining the clustering
technology used by the engine.
Further reading about clustering
search engines
Calishain, T. (2002). Clustering with search engines.
Law Library Resource Exchange. LLRx.com. Retrieved July
3, 2002 from http://www.llrx.com/features/clusteringsearch.htm.
Calishain, T. (2002). Clustering with search engines
Part 2. Law Library Resource Exchange. LLRx.com. Retrieved
July 3, 2002 from http://www.llrx.com/features/clusteringsearch2.htm.
Sherman, C. (2001). Search Day – WiseNut , the
Google killer? Nah. SearchEngineWatch.com. Retrieved
July 3, 2003 from http://www.searchenginewatch.com/searchday/article.php/2157691.
Holly Gunn is the teacher-librarian at Sackville
High School, Nova Scotia. She can be reached at hgunn@accesscable.net.
|
