372 ◾ Jon Atle Gulla et al.
<owl:Class rdf:ID=”DegreeCourseSchema”>
<rdfs:subClassOf>
<owl:Restriction>
<owl:onProperty rdf:resource=”#vtStart”/>
<owl:hasValue rdf:datatype=”&xsd;date”>
1990-01-01
</owl:hasValue>
</owl:Restriction>
...
Since OWL does not allow such time stamps to be added to relations (data proper-
ties or object properties), their approach cannot be used to model evolving ontol-
ogies in general. Van Atteveldt et al. (2008) solve this problem by modeling all
properties as time-stamped classes, leading to an ontology that is problematic for
reasoning and structural clarity. A more conventional approach would be to model
the time-stamped structure as a separate metamodel or temporal ontology structure
from which snapshot ontologies for particular time periods may be extracted.
A concept is normally understood with reference to a term, an intension, and
an extension. e term, also known as a symbol or signier, is a word used in the
real world to refer to the concept. e intension, often called reference or signied,
is the meaning or sense of the concept, and the extension or referent is the set of
real-world objects to which the concept applies. Whereas the intension of car is
our understanding of the concept, the extension is the set of all imaginable and
unimaginable cars. Generally, terminological changes fall into three distinct cat-
egories (see Figure13.18).
Change of term—A particular term is gradually replaced by another term
that has the same basic meaning. We consider them synonyms, even though they
tend to be used in slightly dierent periods. Although we prefer the car term
to automobiles, the automobiles term was used originally for this concept. In an
Term T
0
Extension E
0
Intension I
0
I
1
E
1
T
1
Concept C
0
C
1
Newtermfor
sameconcept
Content of term
changes
New concept(s)
replaces oldone(s)
“Automobile
vs.
car”
“Phone (fromold
Telephonesto
smartphones)”
“Floppy disk
replaced
byCD”
Figure 13.18 Terminological changes over time.
Semantics and Search ◾ 373
ontology, these variants are modeled as time-stamped synonyms or term variants
of classes.
Change of term content—A concept may change semantically, even though
the term referring to the concept stays the same. Both its intension and extension
may be aected by this change, which is often the result of new technology or of
deeper insight into the matter. e phone term has been used for more than 100
years but our understanding of phones and their relationships to other concepts
have changed signicantly. Ontologically, OWL properties and subclass structures
must be modeled as time-dependent.
Change of concept—e most fundamental change appears when some con-
cepts go out of fashion and other related ones gradually take their places. ere
need not be a 1:1 correspondence between the concepts, and they usually dier
somewhat in terms, intensions, and extensions. An example is the replacement of
oppy disks by CDs for the common purpose of storing computer data. In the
ontology, oppy disks and CDs have separate structures, but the structures will
show certain functional commonalities.
For a semantic search application, it is not sucient to know whether certain
concepts are relevant at a particular time. e need is to know how concepts evolve
and relate to new concepts over time, allowing us to translate the original search
terms into terms more useful for the desired period. Conceptually, searches in
evolving domains involve three basic steps:
1. Query disambiguation with respect to user’s intended time frame [t0′, t0″]. is
step makes use of a snapshot of the ontology and is similar to semantic disambigu-
ation discussed above.
2. Time-dependent query mapping. e original query is translated into a set of
time-dependent queries that for each time period is the best semantic replication
of the original query.
3. Semantic search within time periods. Each time period can be compared to a sep-
arate language in a multilingual search application. e time-dependent queries
can be dealt with using the semantic approaches above and snapshot ontologies for
the indicated time periods.
e critical part of the search process is the temporal mapping of queries. Given a
query Q
0
for time period [t
0
′, t
0
″], the system needs to propose a set of new queries,
Q
1
[t
1
′, t
1
″’] …, Q
n
[t
n
′, t
n
″], in which the aggregate of these time periods corresponds
to the original time period [t
0
′, t
0
″], and each Q
i
is the query semantically most
similar to Q
0
for the period indicated. Q
0
contains terms referring to concepts at
the user’s reference time. In Q
i
, concepts C
0
in Q
0
that are not relevant in [t
i
′, t
i
″]
are replaced with semantically close and time-relevant concept(s) C
i
. is includes
access to C
i
’s properties and structures for semantic query processing. In case only
the reference T
0
is irrelevant and the underlying concept is valid, T
0
is replaced with
time-relevant T
i
as a reference to C
0
in the query.
374 ◾ Jon Atle Gulla et al.
Translating a query Q
0
to time-relevant queries Q
1
, … Q
n
requires that all ontol-
ogy elements be time stamped, and relevant semantic links between time periods,
be established for related concepts and terms. is calls for new approaches to
ontology evolution and management that enable us to produce snapshot ontologies
for particular points of time and represent the semantic relationships between these
ontologies. Due to the complexities of such approaches, evolving ontologies have so
far not been exploited in semantic search applications.
13.9 Conclusions
Semantic search applications are based on the same general search architecture
found in popular web search engines like Google and Yahoo! Some research proto-
types employ only semantic components; most systems adopt semantic approaches
only for parts of their architectures and combine them with a standard syntac-
tic search machinery. Certain systems use semantic indices instead of traditional
inverted les, these include: systems that dene semantic query languages or map
standard queries to semantic representations, systems that present results in terms
of semantic structures and summaries, and systems that oer semantic navigation
of result sets. Often the semantic part is realized as a semantic layer on top of a
conventional search engine core.
To be classied as a semantic search application, a system must provide an
explicit semantic representation of categories, words, or texts. Early semantic appli-
cations often dened their own formal languages for this purpose. Current research
emphasizes the use of standard semantic ontology languages like RDF(S) and
OWL. Although ontologies have proven useful in semantic search applications,
we also see their limitations for temporal searches or searches in terminologically
unstable domains. To date, we have no standardized approach to modeling termi-
nological changes in ontologies, and this complicates their use in domains with
evolving terminologies.
In recent years, we witnessed a number of new commercial search applications
that claim to oer semantic search functionality. ese include Hakia, SenseBot,
Powerset, DeepDyve, Cognition, and Kosmix, among others. While their
approaches vary, they tend to rely on statistical NLP and limited use of semantic
lexical resources. Few attempts have been made to dene domain knowledge or
ontologies, and we have omitted these systems from our analysis of semantic search
applications.
A particular case that was not analyzed in this chapter is the Swoogle semantic
search engine that retrieves and even reasons about formal semantic representations.
Swoogle, however, is not a general purpose search application and is restricted to
retrieving ontologies or les with embedded RDF content on the Internet. Since
Swoogle cannot deal with documents that are not dened semantically in the rst
place, it does not help us organize and manage the vast amounts of information
Semantics and Search ◾ 375
already available on the Internet or in enterprises. Semantic document structures
are also assumed by the OWLIR system in Shah et al. (2002).
In spite of all these commercial systems, we see few real semantic search applica-
tions in the marketplace. e scalability problems of ontology-driven applications
may partly explain the current lack of commercial systems. e most successful
applications are built on top of standard search engines to take advantage of well
tested and scalable search architectures. However, certain other concerns must be
addressed in more detail for semantic search applications to become more widely
used. As semantic search applications rest on semantic domain representations
like ontologies and taxonomies, it is often dicult and/or expensive to employ
the applications in new domains. Ontology engineering is tedious and costly, and
very few freely available ontologies can be plugged into general semantic search
applications. Also, the usability of many semantic search applications has not
been given enough attention, and many applications require users to have inti-
mate knowledge of various formal representation languages. When these issues
have been appropriately resolved, we may see more search applications that truly
allow computers to store, retrieve, and understand information the way humans
would.
References
Amaral, C. et al. Design and implementation of a semantic search engine for Portuguese. In
Proceedings of Fourth International Conference on Language Resources and Evaluation,
Lisbon, 2004, pp. 247–250.
Anyanwu, K., A. Maduko, and A. Sheth. SemRank: Ranking complex relationship search
results on the semantic web. In Proceedings of 14th International Conference on World
Wide Web, Chiba, Japan, 2005, ACM Press, pp. 117–127.
Athanasis, N., V. Christophides, and D. Kotzinos. Generating on-the-y queries for the
semantic web: ICS-FORTH graphical RQL interface (GRQL). InProceedings of ird
International Semantic Web Conference, Hiroshima, 2004, pp. 486–501.
Ben-Yitzhak, O., N. Golbandi, N. Har’El, et al. Beyond basic faceted search. In Proceedings
of International Conference on Web Search and Web Data Mining, Palo Alto, CA, 2008,
pp. 33–44.
Berners-Lee, T. et al. Tabulator: Exploring and analyzing linked data on the Semantic Web.
In Proceedings of ird International Semantic Web User Interaction Workshop, Athens,
GA, 2006.
Bernstein, A. et al., Ginseng: A guided input natural language search engine for querying
ontologies. In Proceedings of JENA User Conference, Bristol, UK, 2006.
Bonino, D., F. Corno, and L. Farinetti. DOSE: A distributed open semantic elaboration
platform. In Proceedings of 15th IEEE International Conference on Tools with Articial
Intelligence, Sacramento, CA, 2003.
Bonino, D. et al. Ontology-driven semantic search. WSEAS Transactions on Information
Science and Applications, 1: 1597–1605, 2004.
Budanitsky, A. and G. Hirst. Evaluating WordNet-based measures of lexical semantic relat-
edness. Computational Linguistics, 32: 13–47, 2006.
376 ◾ Jon Atle Gulla et al.
Burkhardt, F., J.A. Gulla, J. Liu, et al. Semi automatic ontology engineering in business
applications. In Proceedings of Workshop on Applications of Semantic Technologies, 2008.
Burton-Jones, A. et al. A heuristic-based methodology for semantic augmentation of user
queries on the Web. In Proceedings of 22nd International Conference on Conceptual
Modeling, Chicago, 2003, pp. 476–489.
Calvanese, D. et al. What to ask a peer: Ontology-based query reformulation. In Proceedings of
Ninth International Conference on Principles of Knowledge Representation and Reasoning,
Whistler, BC, 2004, AAAI Press, pp. 469–478.
Celino, I. et al. Squiggle: A semantic search engine for indexing and retrieval of multi-
media content. In Proceedings of First International Workshop on Semantic-Enhanced
Multimedia Presentation Systems, Athens, 2006, pp. 20–34.
Cohen, S. et al. XSEarch: A semantic search engine for XML. In Proceedings of 29th
International Conference on Very Large Data Bases, Berlin, 2003, pp. 45–56.
Davies, J., R. Weeks, and U. Krohn. QuizRDF: Search technology for the Semantic Web. In
Proceedings of 37th International Conference on System Sciences, Hawaii, IEEE Computer
Society, 2004.
Duke, A., T. Glover, and J. Davies. Squirrel: An advanced semantic search and browse facility. In
Proceedings of Fourth European Semantic Web Conference, Innsbruck, 2007, pp. 341–355.
Eder, J. and C. Koncilia. Modelling changes in ontologies. On the Move to Meaningful Internet
Systems Workshops, Lecture Notes in Computer Science Series 3292, 2004, pp. 662–673.
Fang, W.D. et al. Toward a semantic search engine based on ontologies. In Proceedings of
International Conference on Machine Learning and Cybernetics, 2005.
Finin, T. et al. Information retrieval and the Semantic Web. In Proceedings of 38th Annual
International Conference on System Sciences, Hawaii, 2005, IEEE Computer Society.
Formica, A.,M. Missiko,E. Pourabbas, et al. Weighted ontology for semantic search.OTM
Conferences, 2: 1289–1303, 2008.
García, E. and M.Á. Sicilia. Designing ontology-based interactive information retrieval inter-
faces. In Proceedings of Workshop on Human Computer Interface for Semantic Web and
Web Applications, Lecture Notes in Computer Science 2889, 2003, pp 152–165.
Glover, E.J. et al. Web search your way. Communications of ACM, 44: 97–102, 2001.
Grandi, F., F. Mandreoli, R. Martoglia, et al. Enhanced access to e-government services:
temporal and semantics-aware retrieval of norms. In Semantic Web Applications and
Perspectives: Proceedings of Second Italian Semantic Web Workshop, University of Trento,
Trento, Italy, December 2005.
Guarino, N., C. Masolo, and G. Vetere. OntoSeek: Content-based access to the Web. IEEE
Intelligent Systems, 14: 70–80, 1999.
Guha, R., R. McCool, and E. Miller. Semantic search. In Proceedings of Twelfth International
Conference on World Wide Web, Budapest, 2003.
Guo, L., F. Shao, C. Botev, et al. XRANK: Ranked keyword search over XML documents.
SIGMOD, 2003.
Hein, J. and J. Hendler. Searching the web with SHOE. In Articial Intelligence for Web
Search: Papers from AAAI Workshop, 2000, AAAI Press. pp. 35–40.
Hildebrand, M., J. von Ossenbruggen, and L. Hardman. An analysis of search-based user
interaction on the Semantic Web. INS-E0706, 2007.
Hogan, A., A. Harth, and S. Decker, ReConRank: A scalable ranking method for semantic
web data with context. In Proceedings of Second Workshop on Scalable Semantic Web
Knowledge-Based Systems, 2006.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.