Docstoc

metasearch

Document Sample
metasearch Powered By Docstoc
					Meta Search Engines
 Advanced Information Retrieval
        May 22, 2006

         Ms. Cameron
Outline

 What is a meta search engine (MSE)?
 How meta searching works
 Advantages and disadvantages
 Seven rules for good MSEs
 Demonstration of our six favourite MSEs




                                            2
 What is a Meta Search Engine?

―Meta search engines are powerful tools that
search multiple search engines simultaneously.
 Unlike crawler-based search engines such as
Google, AllTheWeb, AltaVista and others, meta
   search engines generally do not build and
 maintain their own web indexes. Instead, they
use the indexes built by others, aggregating and
 often post-processing results in unique ways.‖

   --Chris Sherman (SearchEngineWatch)

                                                   3
  How Meta Searching Works

 You type in a query
 The MSE translates your query
 The MSE simultaneously searches the
  indexes of other search engines
 The MSE compiles the results and
  displays them in a variety of ways:
     Source
     Relevance (as one list with duplicates removed)
                                                        4
  Advantages of MSEs

 Searches multiple search engines
  simultaneously, saving time
 Ability to easily compare results from
  different search engines
 Broader scope
 Searches broad and shallow, presenting an
  overview of the query

                                          5
Disadvantages of MSEs

 Difficult to conduct a complex search
  because each search engine has its own
  search features
 Sacrifice comprehensiveness for speed
 Search engines not always easy to identify
 Google not always used




                                          6
  ―The rationale for using meta-search engines is to
  save time…This idea seems to be better in theory
than in practice, however, since carefully constructed
  searches of the major search engines often return
 more relevant results than those delivered by many
                meta-search engines.‖
                    --Holly Gunn


                                                    7
Interface

 ―If a site is designed and structured well,
  folks trust it.‖ (Ibid.)
 Because Google is the unquestioned
  leader in the search engine market, their
  basic search interface has been adopted
  many other companies.



                                                8
Comprehensiveness

 Meta search engines search the indexes
  of other search engines
 Their results are only as good as those
  of the engines searched
 Meta search engines should send
  queries to the most comprehensive
  search engines, particularly Google.


                                            9
Powerful Searching

Good meta search engines should support:
 Boolean searching
 Field searching (in title, in URL, etc.)
 Exact phrase matching (― ‖)
 Useful limits
     Last update
     Language
     File type

                                         10
Value-Added Features
 Good Value-Added Features cont.:
     Suggestions for searching, such as
      ―Related Searches‖
     Search areas such as Shopping, Images,
      News, etc.
     The ability to refine searches
     Directories
     Filtering
     Personalization/free registration

                                               11
        Meta Search Engines We Like

        Dogpile*                        ProFusion
        Fazzle                          SurfWax*
        Ixquick                         Turbo10
        Kartoo*                         Vivisimo*
        Mamma*                          ZapMeta




* SearchEngineWatch 2003 Award Winner for Best Meta Search Engines

                                                                     12
ZapMeta
www.zapmeta.com

 Very new; officially released September
  19, 2003
 Searches: AOL, Yahoo!, AllTheWeb,
  WiseNut, Ask Jeeves, Yahoo Web
 Developed by a team of college students
 Offers directories using the Open
  Directory Project and a product search
  powered by PriceGrabber.

                                            13
ZapMeta features two comprehensive HELP menus; one instructs
users in search features, the other orients the user to ZapMeta and
Metasearching in general.



                                                                      14
In addition to
Boolean operators,
Zapmeta also allows
users to use a
proximity operator,
wildcard, and
nesting.
These advanced
search functions are
rarely seen in either
conventional or
metasearch engines.

However…



                        15
Here is the
exact search
that is
demonstrated
in the Help
Menu




               16
Unfortunately, the
Help Menu is out-of-
date or inaccurate,
because the
proximity operator
NEAR has been
searched as a
keyword.




                       17
One of the most
useful features
offered by ZapMeta
is the ―Results
Snapshot‖
The default for this
feature is ―Off‖, but
when turned on…




                        18
A small preview
window gives the
user an idea of
what they will see if
they click a link.
This is quicker and
easier than using
the ―Quick View‖,
although it does not
work for every link.




                        19
The results
page clearly
shows which
results are
sponsored
links.




               20
             Users may also
             specify which
             search engines to
             use, and can even
             select a ―timeout‖
The          period of anywhere
Advanced     from one to 20
search       seconds. This
allows       allows searchers
users to     more control over
control      whether their search
many         is ―quick and dirty‖
aspects of   or more in-depth.
their
search.
             However, note that
             Google is not one of
             the search engines
             used.                  21
ZapMeta

Strengths
 Paid links are clearly labeled
 Users can select which search engines
  are used
 Snapshot feature is useful
 Users may sort results a variety of ways




                                             22
ZapMeta

Weaknesses
 Does not send queries to Google
 Claims to support proximity operators,
  but does not
 Inaccurate Help
 Snapshots not available for every result




                                             23
Fazzle
www.fazzle.com

 A relative newcomer; used to be
  SearchOnline
 Searches: Altavista, Teoma, WiseNut,
  LookSmart, Yahoo, Thunderstone, MSN,
  Dmoz, and Netscape
 A little rough around the edges, but it
  has potential


                                        24
Fazzle‘s home page
provides quick links in
the upper left corner,
links to your top
favourites, two
different help links,
and a news link.
                          25
                                     Users can select what
                                     kind of search they
                                     want by selecting
                                     options from the drop-
                                     down menu
Users can use the first drop-down
menu to choose where they want
to search, then select a sub-
category for that selection in the
second drop-down menu




                                                              26
Fazzle‘s help section is very Windows-like,
with Contents, Index, and Search tabs on the
left, and the content window on right. While
the content is fairly comprehensive, much of
it is out of date.




                                               27
Search results are automatically ranked by
popularity, with rankings clearly displayed. The
engines that were searched are listed along the
top and are linked so you can view just the results
from that search engine. For each search result,
Fazzle displays which engine retrieved the result,
and what its ranking in that engine was.              28
This corner is where you can manipulate the
results. Using the drop-down menus, you can
choose how you want your results grouped
and sorted. You can use the checkboxes next
to each result, then click the Save link to put
them in your Fazzle Favorites. You can click
the Results link to e-mail your selected results,
or e-mail all results.




                                                    29
By clicking the preview link
next to each result, you can
open a preview pane to take
a quick look at the page. You
can maximize the preview
pane to make it larger, or you
can choose to open the page
into the full browser window.
                           30
In the Search Options
window, you can choose
which search engines
you want Fazzle to
search. You can also
change the default
settings for search
results display. Because
Fazzle encourages you
to create an account and
log in, these settings can
be accessed from any
computer.
                             31
Fazzle provides you with
a favourites folder in
which you can also create
subfolders. When you log
in to Fazzle, these
favourites are available
from any computer.


                            32
Fazzle

Strengths
 Allows you to log in and access your saved
  settings and favourites from any computer
 Multiple ways to manipulate search results


Weaknesses
 Doesn‘t search Google
 Out-of-date help file
                                         33
Dogpile
www.dogpile.com

 2003 first place winner for Best Meta
  Search Engine at SearchEngineWatch
 Most popular meta search engine
 Searches: Google, Yahoo, Ask Jeeves,
  About, LookSmart, Overture, FindWhat
 Ability to search five different areas: web,
  images, audio, video, news


                                            34
Dogpile‘s main page is ad-free and is
fairly simple with a good layout, yet it still
provides all the important links a user will
need, such as Advanced Search and
About Dogpile.                                   35
The Advanced Search
page provides separate
fields to break up different
parts of your search.
Dogpile doesn‘t support
Boolean searching. It treats
Boolean words as
keywords.




                           36
The Advanced Search
page also provides
options for manipulating
your search, such as
date limiting and domain
filtering.




                      37
Dogpile only
provides a list of
results and does
not include extra
information.
There are two
sorting options:
relevance and
search engine.
Sorting by
search engine is
the only way to
see which
results came
from which
engine.
           38
                                  Sponsored results aren‘t
                                  separated from the rest of the
Dogpile provides a list of        results, but they are labeled.
alternative search terms and/or
phrases to help you revise
                                                            39
your search
The Tools & Tips page
provides links to helpful
and interesting resources.
You can learn about meta
searching, learn searching
tricks, and see what people
are searching for on
Dogpile.

                       40
SearchSpy lets
you view a
scrolling list of
real-time
searches being
conducted on
Dogpile. The
searches are
hyperlinked so
that you can
view the results
for yourself.




                    41
Dogpile is the only meta search
engine we encountered that has its
own mascot. His name is Arfie and
he has his own bio on the site.      42
Dogpile

Strengths
 Easy to use because it doesn‘t have a lot of bells
  and whistles
 Clean, streamlined, ad-free interface


Weaknesses
 Can‘t pick and choose which search engines to
  search
 Doesn‘t support Boolean, truncation or wildcards

                                                 43
Kartoo
www.kartoo.com

 Meta Search Engine from France
 Created by a man called Laurent
  Baleydier
 It uses Flash Player to display search
  results in interactive maps.




                                           44
Kartoo Features

 Truncation: *
 Phrase Searching using quotes
 Natural Language: ?
 Boolean: AND, OR, NOT, NEAR
 Limits: Domain, Site, Title, Text, Link and
  Like, Host and URL.



                                            45
Kartoo Features, cont.

 Exporting options
     Searches and maps can be saved, printed,
      and e-mailed
     Searching preferences can be saved
     Parental filters are optional




                                             46
Kartoo‘s Transparency

 Very transparent. All of user‘s questions
  can be easily answered with a little digging
 A list of the search engines are provided
  from the options menu above the map
 Paid sponsors are clearly marked




                                              47
How does Kartoo work?

 User enters search terms
 Kartoo sends queries to several search
  engines
 Kartoo classifies the results by theme
  and relevance and then displays them
  cartographically



                                           48
How do you read these interactive
maps?
 The maps are created in such a way that
  a user can quickly learn how to use them
  by simply moving the mouse around the
  map and watching the different changes
  that occur.




                                         49
Navigating the Map

 Results are displayed with open book icons.
 Relevant results are closest to the middle and
  are larger books.
 Less relevant results take up the perimeter and
  are smaller. Open pages in a book means there
  are more web pages from the same site.
 The user can view results and simultaneously
  refine their search by simply maneuvering the
  mouse over the map. Nothing is changed until
  the user clicks the mouse.
                                                   50
Previewing results using the
mouse
 User can read a small description of
  each result by guiding the mouse over it.
 To the left of the map is a list of the top
  web sites for that search, moving the
  mouse over these results causes the
  corresponding sites to light up on the
  map and a brief description of the site
  appears below the list.

                                            51
Refining the search as you go

 As the user clicks on different results, the
  search terms in the search box located above
  the map change accordingly
 This allows a user to easily narrow their search
  according to the search engine‘s suggestions
 Under the list of sites is list of subject areas in
  quotes which are more suggested topics for
  your particular search



                                                    52
Thematic Display

 Results are clustered by themes or
  subject areas.
 Relationships are shown through
  connecting lines between results.
     Mousing over a result highlights the
      relationship and turns the line green




                                              53
What does Kartoo do best?

 Kartoo meets the seven criteria outlined
  previously, but its unique interface and
  ability to display results both visually and
  thematically is its most outstanding and
  unique feature.




                                             54
Why a map instead of a list?

 A list of results can only portray relevance
  linearly, usually from most to least.
 A map allows a user to visually compare
  the relevance of results (spatially and
  through the use of large and small icons).
 The relationships between results can be
  visually portrayed to the user.


                                            55
Themes are represented

 The classification process and visual
  display allows the themes within a
  search to be articulated to the user in a
  visual format.




                                              56
Marketing Manager of Kartoo
rationalizes the approach
 ―Instead of pages of results listing sites, Dos
  Santos says his company pictured more of a
  road map so they could help users navigate the
  results. ‗We imagined cities instead of Web
  sites and instead of roads, common words
  linking destination sites. The user can zoom
  into thematic areas to improve the search…the
  search engine divides the results into logical
  categories, making it easier for users to zoom
  into the categories which make the most
  sense‘‖ (Miller 2004, 34).
                                                    57
Problems/Weaknesses

 Commercial/Business oriented
 Takes some time for the user to orient
  themselves to the map format



www.kartoo.com



                                           58
SurfWax
www.surfwax.com

 Flexible search tool that features a
  number of unique search and display
  options
      Some options only available to registered
       users (basic registration is free)
 California-based company founded in
  1994
 Honorable mention for Best Meta Search
  Engine by Search Engine Watch in 2002
                                                   59
   SurfWax               Registered users can
                          also select from
                          hundreds of search
 Primary information     engines, databases, and
  sources:                websites to create
      Yahoo              personalized
      WiseNut            ―SearchSets‖
      Yahoo News
      CNN               Power searching:
      AOL                  Basic prefix operators
      LookSmart           (+ or – )
      MSN                  Exact phrase matches
                           (―. . .‖)

                                                      60
Home page
provides pull-down
menu options for
selecting the
number of results to
display as well as
the sorting method
(i.e. by relevance,
alphabetic order, or
source).
                   61
Search results
displayed on left
side of screen
(sources clearly
indicated).
Sponsored links
displayed on right
side. ―FocusWords‖
are noted above
results display.


                62
By clicking on a FocusWord, users are directed to SurfWax‘s online
thesaurus. Broader and narrower categories of related terms are listed as
active links.


                                                                            63
Back on the results page, users can
click on the magnifying glass icons
next to result titles to display
―SiteSnap‖ information (content
analysis and highlighted keywords).




                                64
SiteSnaps also highlight page
elements (Key Points as well as
FocusWords). Registered users
can save results to their
personalized ―InfoCubbies.‖




                            65
Registered
users also
have the
option to
select their
search
preferences
(number of
results,
display
format,
etc.).

          66
 Surfwax
Strengths
 Excellent value-added features available to
   registered users
 Good tool for ongoing research
Weaknesses
 Field searches and other common limiters not
   available
 Information overload?
     Cluttered display screen
     Learning how to take advantage of numerous
      search features may be time-consuming
                                                   67
Vivisimo
www.vivisimo.com

 ―Clustering‖ search technology developed by
  researchers at the Computer Science
  Department of the Carnegie Mellon University
 Search results instantly organized into
  hierarchical subject categories
 Company founded in 2000
 Won Best Meta Search Engine award from
  Search Engine Watch in 2002; second place in
  2003


                                                68
Vivisimo

 Primary information    Search features:
  sources:                    Boolean operators
      Looksmart               (AND/+, OR, NOT/-)
      Lycos                  Exact phrase
      MSN                     matches (―. . .‖)
      Overture               Limiters (image:,
                               title:, domain:, host:,
      Wisenut
                               etc.)




                                                     69
Vivisimo

 ―Vivisimo is a good choice if you want to
  use a meta search engine, but a little
  digging beneath the surface reveals
  some absolute gems for searching
  specialized information sources‖
     (Sherman, ―Power Searching with
     Vivisimo,‖ 2003)



                                              70
Home page features a drop-
down menu that provides a
variety of search options
(including a number of
specialized news categories
and sites).
                              71
Advanced search
page also allows
users to select
specific
language, filter,
and display
options. Easy-to-
use ―Help‖
feature is always
available.




                    72
Results are
―clustered‖ into
related subject
groups.
Information
sources (paid
sponsors or
other search
engines) are
clearly identified.
               73
Users can
search within
results
(keywords
highlighted
in yellow).
The results
page also
offers a
number of
viewing
options (new
window,
frame, or
preview)

         74
Vivisimo

Strengths
 Easy-to-use basic and advanced search
   interfaces
 Excellent search and display options
 ―Clustering‖ technology provides broad
   overview of relevant results (from most general
   to most specific)
Weaknesses
 Commercial ―feel‖?
      Originally launched as a product demo

                                                 75
Vivisimo introduces Clusty
http://clusty.com

 In September 2004, Vivisimo launched Clusty
 ―Despite the awkward name . . . Clusty is an
  elegant new search tool that takes underground
  favorite Vivisimo to a new level. In addition to
  presenting both standard web search results
  and Vivisimo's dynamic clusters that
  automatically categorize results, Clusty draws
  on several new sources of information.‖
  (Sherman and Price, ―Reducing Information
  Overkill, 2004)
 Additional search categories include News,
  Images, Shopping, Encyclopedia, and Blogs
                                                 76
Conclusion

 Some meta search engines are better
  than others
 Features and technology change rapidly;
  check back often!
 It is a matter of preference and choice
 You have to experiment to determine
  which features work best for you


                                        77
                                        References
Arnold, Stephen E. ―Vivisimo: Clustering Delivers Information Overlook.‖ (5 May 2003).
    http://www.arnoldit.com/articles/iwr_vivisMay2003.html

Bazak, Daniel. ―The Meta Search Engines: A Web Searcher‘s Best Friend.‖ (10 October 2002).
   http://evolt.org/article/rdf/12/41694/index.html

Carty, Susan. ―Search Engines and Meta Search Engines.‖ Information Searcher 14, no. 1 (2003): 4-9.

Fifield, Craig. ―Effective Search Engine Design.‖ (7 November 2002).
     http://searchenginewatch.com/searchday/article.php/2161151

Glover, Eric J. et. al. ―Web Search—Your Way.‖ Communications of the ACM 44, no. 12 (2001): 97-102.

Gunn, Holly. ―Searching with Meta-Search Engines.‖ Teacher Librarian 31, no. 3 (February 2004): 50.

Gunn, Holly. ―Searching the Web with Free Visual Search Engines.‖ Teacher Librarian 31, no. 5 (2004): 51.

Luther, Judy. ―Trumping Google? Metasearching‘s Promise.‖ Library Journal 128, no. 16 (1 October 2003): 36-39.

Miller, Ron. ―Get the Picture: visualizing the future of search.‖ EContent 24, no. 4 (2004): 30-32.

Ojala, Marydee. ―Search, Navigate, Find, Personalize.‖ Information Today (January 2004): 33.

Repmen, Judi and Randal D. Carlson. ―Surviving the Storm: Using Metasearch Engines Effectively.‖ Computers in
   Libraries (May 1999): 50-55.
                                                                                                     78
                            References, cont.
Sherman, Chris. ―Dogpile Sports a Fetching New Look.‖ (2 September 2003).
   http://searchenginewatch.com/links/article.php/3070811

_____. ―Hang Ten with Surfwax Metasearch.‖ (10 July 2001).
   http://searchenginewatch.com/searchday/article.php/2157271

______. ―Meta Search Engine Week.‖ (16 September 2002).
   http://searchenginewatch.com/searchday/article.php/2160771

_____. ―Metacrawlers and Metasearch Engines.‖ (15 March 2004)
   http://searchenginewatch.com/links/article.php/2156241

_____. ―Power Searching with Vivisimo.‖ (8 July 2003).
   http://searchenginewatch.com/searchday/article.php/2226841

_____. ―ZapMeta: A Promising New Meta Search Engine.‖ (26 February 2004).
   http://searchenginewatch.com/searchday/article.php/3317991

Sherman, Chris, and Gary Price. ―Reducing Information Overkill.‖ (30 September 2004).
   http://searchenginewatch.com/searchday/article.php/3415071

Sullivan, Danny. ―Meta Search or Meta Ads?‖ (23 May 2001).
     http://searchenginewatch.com/sereport/article.php/2163821

Sullivan, Danny, and Chris Sherman. ―4th Annual Search Engine Watch Awards.‖ (6 February 2004).
     http://searchenginewatch.com/awards/article.php/3309841#meta

Zhang, Jin. ―Meta-search-engine feature analysis.‖ Online Information Review 27, no. 6 (2003): 433-41. 79

				
DOCUMENT INFO