|
|
|
Search Results Clustering Demystified
lustering may mean to have are coded by programmers using as
two or more computer basis the users' preference on
systems working together or what they want to see on
multiple servers linked together clustered documents. Clusters are
for the purpose of handling presented using the style of
variable workloads as well as to folders and sub-folders.
provide continued operation in
case one fails. It may also refer When a search engine provides
to data clustering which is a millions of results for a
technique used for data analysis particular query, the searcher
by dividing a data set into can either sift through the
subsets whose elements share endless pages of results or
common traits. Search result depend on the search engine's
clustering aims to change the way judgment as to the most relevant
people search online by results. Neither can ensure that
organizing search result into the targeted information can be
folders that group similar items accessed as it may remain buried
together. under pages of results or it may
not meet the search engine's
Why Clustering is Needed criteria. In the same way that
all other things are clustered or
The use of the vast information organized, the world of web
available online cannot be searching would be more useful
maximized unless an effective once given the benefit of
means of organizing it can be organized search results.
provided. Clustering engines put
search results together based on Clustering engines automatically
textual and linguistic cluster results into categories
similarity. This basic similarity that have been intelligently
is supported by heuristics which selected from words and phrases
contained in search results. found on the tenth page to be
Categories are intended to reach just a click away. Related items
human-level accuracy and to offer can also be viewed together
hierarchical drill doom without much effort. It even
capability in a familiar reveals unexpected relationships
folder-style interface. between words, ideas and
Mind-numbing lists need not be concepts.
scrolled through or ignored as
the main themes are viewed in the A good cluster is considered such
first 300 - 500 results right on if it possesses a readable
the first page. A quick overview description. It should be able to
of the types of information assist in narrowing down a search
available on a particular topic to find exact results. A
is made available so that the clustering engine queries
area of interest can be multiple search engines and
immediately put into focus. combines the results to be
clustered and displayed on one
With the great improvement of screen. Each result list comes
search engines' capability to with information regarding the
return a large number of relevant total number of results clustered
results, it became more difficult and retrieved. The clustering
to navigate meaningfully through engine's own heuristics shall
all the results. A typical determine the pages to be
searcher does not take the time favored. Search engines sometimes
to view results beyond the first return multiple copies of the
page which makes it very probable same page with slightly different
to miss results that would have URLs but this is minimized in
been relevant and useful to search result clustering. This is
his/her search or query. Clusters because clustering engines does
make it possible for results not reproduce results with
similar descriptions. Clusters WiseGuide. Some results would
are specific enough that repeated have subtopics which will show
documents are very rare. Some are underneath the clustered results.
able to offer advanced search A link can be found next to each
features which allows searchers of the clustered results whose
to specify which sources should keywords can be used to run
be searched, the number of another search. A different set
results desired, allowable of clustered results shall be
waiting time, the desired produced in addition to the web
language to be used and the page results. This search engine
filtering out of offensive has been bought by LookSmart.
contents.
Teoma has been dubbed as the
Search Engines that Clusters "Google Killer" due to its very
interesting clustering
Google Sets do not provide technology. A single search run
results but rather helps in will produce four sets of
finding similar terms to the ones results. Those found at the top
entered. This allows the user to left are sponsored results, those
create more complex queries in found at the bottom are website
one area and brainstorm on how to non-sponsored results, those at
put a search together. Google the top right are the suggestions
Sets is Google Labs' clustering for refining the result and those
agent. at the bottom right are link
calculations from experts and
Wisenut is a full-text search enthusiasts. The link collections
engine which provides for related are suitable for general
topics aside from a number of information needs while the
results for any search item suggestions are for more specific
entered. This is called the searches. A click on any would
signal the search to run again that clusters its results. It
where a different set of site provides a very simple front page
results shall be provided. Teoma with search results that are
has been purchased by AskJeeves. organized in groups. The page
design makes it easy to explore
Infonetware.com is more of a several categories without having
demonstration of Infonetware's to "lose your place". Clusty is
Real Term Technology than a the consumer search destination
search engine. The results page powered and owned by Vivisimo. It
is framed where the area on the queries results from Ask, MSN,
left provides topics related to Open Directory, LookSmart,
the search term while the web Gigablast and WiseNut. These
page search results are found on sites were chosen because of
the right frame. It works with their accurate results and quick
full searching. return speeds.
Oingo uses the open Directory Query Server offers several types
Project as its search source. The of search on the left side of the
search results page gives a front page. Each search has more
drop-down list of potential or less the same interface and
meanings. The list of categories all cluster results. Search
in order of relevance to the results are presented in a frame
search can be found beneath it as at the right side of the site.
well as the site results from the
directory itself. It is more Surfwax offers both subscription
useful for general term searches based and free services. A focus
or search terms that are in a link can be seen in the upper
broad category. left corner after a search is
entered. These focus words can be
Vivisimo is a meta-search engine used in addition to the search
term. They are divided into are subfolders provided for broad
narrower or broader categories topics. Search results are listed
and contain generic words and not by order of date.
links to specific people or
places. Clustering search engines break
up several hundred results into
Northern Light News search manageable packages. Suggestions
requires a search to have a are provided so that the use of
certain number of results in information is maximized and the
order to be clustered into search itself a lot easier. A
folders. However, folder listing search query cannot always be
does not provide information specific enough to target the
about the contents of a right information at once.
particular folder although there
About the Author:
http://www.theinternetone.net
Read more articles by: Danny Wirken
Article Source: www.iSnare.com
|
|
