State-of-the-Art: Semantics Acquisition and Crowdsourcing - Semantic Acquisition Games: Harnessing Manpower for Creating Semantics - page 11

2.2.2 Exploratory Search
In 2006, Gary Marchionini coined the term exploratory search and marked the birth
of a new search paradigm, which tried to solve the problems with invisibility of the
Web [ 44 ]. The invisibility problem is often formulated as the inability of the Web
user (searcher) to formulate a search query in a particular domain, because of a lack
of familiarity with domain's jargon and structure (it can also be observed in the above
case of ambiguous query reformulation).
Marchionini states that not all web search tasks are sufficiently solvable by key-
word search, namely learning and information composition tasks. He also states
the problems of inexperienced users new to certain domains, who do not know the
domain jargon and are thus unable to type in proper keywords [ 44 ]. Marchionini
suggests the use of alternative search interfaces that visualize some abstraction of
the domain that user can afterward navigate and filter the content rather by browsing,
then by query formulation. However realization of such interface requires a suitable
semantics: both resource descriptions and domain models.
A very basic example of an exploratory search tool is a tag cloud. A relatively
widespread technique for result filtering visualizes an information space as a collec-
tion of words which characterize its content—the individual resources. The words
are usually displayed to the searcher in different sizes reflecting their weight (e.g.,
importance, frequency) in the corpus. The user is then allowed to review the cloud,
learn about the contents of the otherwise invisible corpus. Moreover, he can interact
with the cloud by selecting its individual words to filter the result set.
Usually, tag clouds operate over keywords assigned to the resources. However,
they could be used with more elaborate semantics too. One example of an search
and navigation application exploiting ontologies in this way is Idea Navigation [ 63 ].
With simply looking textual interface, it directly uses the triplets (instead of tags) to
propose possible options of filtering the result set to users. This allows the searcher
to learn more about the domain he is reviewing.
A much more elaborate example of exploratory search approach is a faceted
browser which heavily relies on semantics. One of such named Factic was cre-
ated by Tvarožek and Bieliková [ 67 , 68 ]. The faceted browser is an information
retrieval application that uses facets (a filter criteria along with their possible values)
as the means of formulating of the search queries (rather then by typed-in keywords).
Faceted browsers are typical in e-shops where the visitor can filter the products by
their parameters (in Factic, the facets are even generated automatically based on
metadata in ontological repository to which it is attached). The provision of such
approach is apparent, the customer can immediately see what is available and he
already has a good overview of the whole information space on the e-shop portal,
because he sees the category structure and also a list of faceted criteria and their
possible values. This, of course, depends on well devised ontology or at least set of
homogeneous resource descriptions.