Trapping Topic-Based Information - Information Trapping: Real-Time Research on the Web - page 114

tr y a clu ster for Your search

perhaps you're having problems developing the perfect query—the exact words you want are
eluding you. in these cases, it's a good idea to veer off the track of regular search engines for a
moment and try something else: clustering search engines.

a clustering search engine “clusters” results into groups. The groups are built around a certain topic.
say you do a search for chips . a clustering search engine might cluster your result into groups like
potato chips, poker chips, wood chips, and chocolate chips. instead of having to review all the search
results at once, you could then choose a group that was most relevant to your search.

clustering search engines can help you igure out query words that are relevant to search. Let's
look at the chip example again. if you're trying to do a query on poker chips, you could peruse
the cluster that groups results about poker chips and see what words appear over and over again.
ask yourself what caused the clustering engine to put all those results together. perhaps those
words are ones you could use to reine your query.

There are many clustering search engines available. here are a few for you to try:

• Clusty (clusty.com) is a service of Vivisimo, one of the irst clustering search engines

available. conducting a search for “chips” on clusty not only produced a variety of clusters,
including “tortilla” and “health programs” (sometimes the reasons for a cluster aren't
immediately obvious!), but it also produced lists of possible matches for chips within search
results (in this case, from Wikipedia pages).

• Mooter (mooter.com) displays lists of search results using a visual layout of the clusters

it inds to match your query (Figure 7.5). once you choose a cluster you want to explore
further, it then provides a more traditional layout, with clusters on the left and search
results on the right.