ITERATIVE XML SEARCH BASED ON DATA
AND ASSOCIATED SEMANTICS
Alda Lopes Gançarski
Institut TELECOM, TELECOM et Management SudParis, CNRS SAMOVAR, 9 rue Charles Fourier, 91011 Évry, France
Pedro Rangel Henriques, Flávio Xavier Ferreira
Department of Informatics, University of Minho, Campus de Gualtar, 4710-057 Braga, Portugal
Keywords: XQuery, iterative XML search, metadata, ontologies, RDF, SPARQL.
Abstract: In a previous work in the context of information retrieval, XQuery was extended with an iterative paradigm.
This extension helps the user getting the desired results from queries. In a related work, XQuery was also
extended to allow the inclusion of SPARQL queries; this is useful when XML documents are associated
with semantic RDF descriptions. However integrating SPARQL in XQuery queries makes the construction
of queries more complex (although more powerful). To leverage this integration, we propose to apply the
iterative paradigm to the ‘SPARQL extension to XQuery’. In the paper this proposal is introduced and
justified and a case study is presented.
1 INTRODUCTION
1
XML information access is done using structured
query languages such as XPath (Berglund et al.,
2007) and XQuery (Boag et al., 2006), the standard
proposed by the W3C. To help the user get the
desired information from his queries, (Gançarski et
al 2006) proposed an iterative search over XML
documents using an extension to XPath. The
iterative paradigm was, then, included in XQuery
(Gançarski and Henriques, 2006).
To improve data processing, the document
collections and Web resources are associated with
semantic descriptions, i.e. metadata. In order to be
able to exchange the semantics of information, one
first needs to agree on how to explicitly model it.
This is usually done using a sophisticated
description in the form of an ontology. An ontology
is a formal explicit specification of a shared
conceptualization. Using an ontology, any kind of
description can be made about a resource.
Ontologies can be used to annotate data with labels
indicating their meaning, thereby making their
semantics explicit and machine-accessible. W3C
1
This research is done in the context of the RESPIRE
project financed by the French ANR-ARA program.
has created the Resource Description Framework
(RDF) (Manola and Miller, 2004), a language for
representing information about resources in the
World Wide Web. RDF Schema (RDFS) is an RDF
extension, which provides the basic elements for
ontologies descriptions. To find information in RDF
descriptions, the SPARQL query language was
defined by the W3C (Prud’hommeaux and Seaborn,
2007).
In (Gançarski and Henriques 2007), they exploit
the use of XML documents together with the
respective semantics to access information, arguing
that both may be interesting to the user and can help
him to find the desired information. For that, they
integrate SPARQL queries into XQuery ones.
In this paper, we propose to apply the iterative
paradigm to the SPARQL extension made to
XQuery. In fact, searching by data and metadata is
more sophisticated than simple data search. Thus,
the user may take advantage from the iterative
paradigm when building his queries, not only in the
XQuery component, but also in the SPARQL one.
Next section introduces the XQuery iterative
model. Section 3 introduces the search based on data
and associated semantics. Then, Section 4 integrates
the iterative paradigm with the semantic search. The
formal definition of the proposed XQuery extension
is made in Section 5. A case study is described in
479
Lopes Gançarski A., Rangel Henriques P. and Xavier Ferreira F. (2008).
ITERATIVE XML SEARCH BASED ON DATA AND ASSOCIATED SEMANTICS.
In Proceedings of the Tenth International Conference on Enterprise Information Systems - DISI, pages 479-484
DOI: 10.5220/0001706504790484
Copyright
c
 SciTePress

Section 6. The article finishes with a brief
conclusion, indicating some future work. For the
sake of simplicity, we will ignore IRIref definitions
in our examples.
2 ITERATIVE SEARCH
The iterative paradigm of query construction is
based on selection operations that consist in
restricting intermediate results to the subset of
elements that satisfies the user. For that, the
mf:select function may be used in location path
expressions (mf from my function). The mf:select
function selects the subset of interesting elements
based on some criteria. While in a filter the set of
elements is selected by intention, with mf:select it is
by extension, i.e. explicitly referring to each
element. This can be interesting when the
specification of the criteria is too complicated (the
user may even not know how to do it) or when it is
more efficient/rapid to directly refer the desired
elements.
Suppose each node is identified by a unique
identifier and consider it as a string of characters.
The input to mf:select is a node and a list of node
identifiers (denoted by “(...)”). The output is the
input node if it is selected (i.e., if its identifier
belongs to the list of identifiers), or an empty
sequence of nodes. For example, suppose the user
wants references made inside interesting articles of
author Kevin. The user can, then, make the
following query:
for $a in /articles/article[author = ``Kevin'']
[mf:select(., ("a4", "a8"))]
return $a//references
In this query, function mf:select selects articles
identified by "a4" and "a8". Symbol “.” refers to
each context node, i.e., each resulting node of the
precedent operation. Thus, mf:select takes each
article being a context node and returns it if it
corresponds to some of the selected items.
3 DATA AND SEMANTICS
SEARCH
When an XML document is associated to semantic
descriptions expressed in RDF, SPARQL queries
may be integrated in XQuery queries. This can be
done by adding a new clause metadata to the for
clause of XQuery. Let us call the extended XQuery
with the metadata clause XQuery+SPARQL. As an
example, consider the following ontology that
includes as concepts, among other things, elements
of an XML document.
Book.xml about “Beings”.
Book.xml#Chapter1 about “Fishes”.
Book.xml#Section11 about “Ocean Fishes”.
Book.xml#Section12 about “River Fishes”.
Book.xml#Chapter2 about “Birds”.
Book.xml#Chapter3 about “Vegetables”.
Book.xml#Chapter4 about “Fruits”.
Men eat “Birds”.
Men eat “Fishes”.
Men eat “Vegetables”.
Men eat “Fruits”.
If the user wants chapters about what men eat, he
can specify the following XQuery+SPARQL query:
1 for $c in
2 doc(“http://.../Book.xml”) /book//chapter
3 metadata $c in
4 SELECT ?c
5 WHERE { Men eat ?o.
6 ?c about ?o. }
7 return $c
In this query, the for clause associates to variable
$c the set of chapters of the document (lines 1 and
2). The metadata clause (line 3) includes a SPARQL
query (line 4 to 6) which selects all the book parts
(stored in variable ?c) that are about beings eaten by
men (stored in variable ?o). This yields the set
{Book.xml#Chapter1, Book.xml#Chapter2,
Book.xml#Chapter3, Book.xml#Chapter4}, stored in
variable ?c. This result is intersected with the set of
elements of the XQuery external query, stored in
variable $c, to get the desired parts of the book.
4 ITERATIVE DATA AND
SEMANTICS SEARCH
The formulation of XQuery+SPARQL queries is
more complex than simple XQuery queries. To
simplify this task to the user, we propose to extend
the iterative paradigm to the SPARQL component of
XQuery+SPARQL. For that, a selection function is
included in SPARQL, associated to the FILTER
clause. Let us see an example query using a filter
and, then, incorporate the selection function on it.
ICEIS 2008 - International Conference on Enterprise Information Systems
480

4.1 FILTER Clause
SPARQL filters restrict solutions to those for which
the filter expression evaluates to true. Filters use
functions to define conditions. Those functions may
come from XPath, XQuery or may be SPARQL
specific. For example, consider the same ontology as
before in Section 3. Suppose the user wants sections
about fishes. He may, then, specify the following
query:
1 for $s in
2 doc(“http://.../Book.xml”) /book//section
3 metadata $s in
4 SELECT ?s
5 WHERE { ?s about ?o.
6 FILTER regex(?o, “Fishes”) }
7 return $c
The regex function (line 6) is SPARQL specific and
allows matching a string with a pattern. In the
example query, the string is stored in variable ?o and
the pattern is the simple string “Fishes”. The result
of the Where clause is the set { Book.xml#Chapter1,
Book.xml#Section11, Book.xml#Section12}. When
making the intersection with the result of the
external XQuery query (which gives sections from
Book.xml), the final result becomes
{Book.xml#Section11, Book.xml#Section12}.
4.2 Select Function
With the proposed iterative paradigm, when a
variable is computed, the user may see its content
and select the subset of interesting values. For that,
we define the msf:select function (msf from my
SPARQL function). Suppose again the ontology of
the previous example in Section 3. If now the user
wants animals eaten by men, he may specify a query
in the following steps:
Step 1.
for $s in
doc(“http://.../Book.xml”) /book//chapter
metadata $s in
SELECT ?s
WHERE { Men eat ?o.
At this point, with the iterative paradigm, the user
may access the intermediate result stored in variable
?o. This result is the set {“Birds”, “Fishes”,
“Vegetables”, “Fruits”}. The user may, then, select
the animals, as specified in the next step.
Step 2.
for $s in
doc(“http://.../Book.xml”) /book//chapter
metadata $s in
SELECT ?s
WHERE
{ Men eat ?o.
FILTER msf:select(?o, (“Birds”, “Fishes”))
?s about ?o.
}
return $s
With the msf:select operation, the content of variable
?o became {“Birds”, “Fishes”}. Thus, the final
result of the query is {Book.xml#Chapter1,
Book.xml#Chapter2}.
5 FORMAL DEFINITION OF
SELECT FUNCTION
The FILTER clause may be associated to user
defined functions, as specified in the following
productions:
[26] Filter ::= 'FILTER' Constraint
[27] Constraint ::= FunctionCall
The msf:select function is defined using the
SPARQL grammar productions corresponding to
user defined functions:
[28] FunctionCall ::= IRIref ArgList
[29] ArgList ::= ( NIL | '(' Expression ( ','
Expression )* ')' )
Here, the name of the function is derived by IRIref.
This symbol allows for complete IRI references or
prefixed names. In our case, we use the prefix msf
and the name select.
The arguments of user defined functions are
represented by the symbol Expression. This is a
general symbol representing from simple literal
strings to complex Boolean expressions. In the
msf:select function, the first occurrence of
Expression derives an RDF term, corresponding to
the content of some variable, and the second one, a
bracketed sequence of RDF terms. This sequence
corresponds to the sequence of selected terms from
an intermediate result. What follows defines the
msf:select function:
xsd:boolean msf:select(RDF term t,
(RDF term)* tSeq)
ITERATIVE XML SEARCH BASED ON DATA AND ASSOCIATED SEMANTICS
481

{
for (i=0; i<length[tSeq]; i++) {
if (sameTerm(t, tSeq[i]))
{ return TRUE; exit; }
}
Return FALSE;
}
The sameTerm SPARQL pre-defined function
returns true if both arguments are equal.
For each term bound to a variable passed as the
first argument of msf:select, this function returns
true if the term exists in the sequence of terms
passed as the second argument; otherwise, it returns
false. Considering the example query of Section 4.2,
variable ?o is bound to the set of terms {“Birds”,
“Fishes”, “Vegetables”, “Fruits”}. This variable is
the first argument of the msf:select function. The
second argument is the sequence (“Fishes”,
“Birds”) choose by the user. So, verifying if each
term stored in variable ?o occurs in the sequence,
the content of ?o becomes {“Birds”, “Fishes”}.
6 CASE STUDY
We intend to experiment our approach with the
resources of the Portuguese Emigration Museum
(Museu da Emigração e das Comunidades - MEC).
MEC is a web-museum that wants to make easily
accessible to the general public the rich cultural
heritage characterizing the Portuguese emigration
phenomenon, and the impress left by the Portuguese
people around the world.
MEC assets (resources) are vast and multifaceted
because the emigration documents and objects come
from the most diversified sources, ranging from
official government records to old newspapers and
photo albums, with the type of documents also
heterogeneous (from official travel reports to local
stories). So it becomes necessary to organize and
categorize all this resources. As such, the
information sources and the resources where
classified using an ontology, presented partially in
Figure 1 (at the end of the paper). Furthermore,
using XML Schema (Fallside and Walmsley, 2004),
we defined, an XML format for each type of
document (the ellipses in Figure 1) - documents can
be seen as the resources being described by the
ontology.
The ontology gives a full categorization of the MEC
resources universe; it shows the relations between
the location of the resources (ex.: district archives),
the information sources (ex.: passport's processes,
almanacs) and the documents themselves (ex.: birth
certificate, passport petition, event record).
Next subsections show different application
scenarios using XQuery+SPARQL over the MEC
resources.
6.1 Using XQuery+SPARQL
A visitor, wishing to explore MEC’s resources, may
be interested in searching for travelling details
registered in Fafe related to the emigrant Antonio
Serra; the details considered below are, the date of
departure, and the destination (target country) .
Travelling information is obtained accessing
passport records; an excerpt of a record of this kind
is as shown in the following simplified:
<passportRecord ...> ...
<name>Antonio Serra</name> ...
<place of="destination">Brasil</place>
<place of="source">Portugal</place>...
<date normDate="1866-11-13">…</date>
</passportRecord>
The XQuery+SPARQL query which yields to
the desired information is:
for $d in doc(“passRec.xml”)/passportRecord
where $d/name = “Antonio Serra”
metadata $d in
select ?d
where {
?d rdf:type Passport_Record .
?d contained_in ?p .
?p rdf:type Municipal_passport_record .
?p acquired_in Fafe .
}
return
<pr> {
$d/passportRecord(place[@of=”destination”]|date)
} </pr>
In the metadata clause of this query, the SPARQL
query searches for documents contained in
municipal passport records (stored in variable ?p)
which were acquired in Fafe.
6.2 Using Iterative XQuery+SPARQL
Suppose, now, the visitor is interested in searching
for images and photos related to some families he
knows. Let us assume a user from the city of Braga.
Family photos can be found in the respective family
album. Instances of family albums are represented
ICEIS 2008 - International Conference on Enterprise Information Systems
482

by identifiers such as “Freitas_Fam_Album” for the
“Freitas” family.
Image and photo documents have a structure/content
similar to the following excerpt:
<image>
<name>Family address of José Freitas</name>...
<city>Braga</city>
<date normDate="1885-01-01">…</date>
<img>http://.../jose_freitas.jpg</img> ...
</image>
Then, the query satisfying the current example is:
for $i in doc(“image1.xml”)/image
where $i/city =”Braga”
metadata $i in
select ?i
where {
?i rdf:type Images_and_photos .
?i contained_in ?f .
?f rdf:type Family_album .
FILTER msf:select(?f, (Freitas_Fam_Album,
Silva_Fam_Album)) .
}
return $i/img
In the metada clause, the list of identifiers of the
family albums found is displayed, as the answer
computed in the third triple pattern of the SPARQL
query (stored in variable ?f). The visitor may, then,
immediately identify those belonging to families he
knows. He can, then, select those albums using the
msf:select function associated to the FILTER clause.
In this query, the user selected albums from
“Freitas” and “Silva” families.
7 CONCLUSIONS
In this paper, we propose to integrate the iterative
paradigm for query construction into the
XQuery+SPARQL semantic querying language. We
believe this can help users to get the desired
information.
We intend to create a prototype processing
environment for the XQuery+SPARQL. We can use
existing XQuery and SPARQL query processors
integrating them with a special editor and result
visualizer. We will, then, test this prototype and
verify the usefulness of this approach using MEC
assets, as described in Section 6.
REFERENCES
Berglund, A., Boag, S., Chamberlin, D., Fernandez, M.,
Kay, M., Robie, J., Siméon, J., 2007. XML Path
Language (XPath) 2.0 W3C Recommendation 23
January 2007, URL: http://www.w3.org/TR/xpath20/.
Boag, S., Chamberlin, D., Fernandez, M., Florescu, D.,
Robie, J., Siméon, J., 2007. XQuery 1.0: An XML
Query Language W3C Recommendation 23 January
2007, http://www.w3.org/TR/xquery/.
Fallside, D. and Walmsley, P., 2004. XML Schema Part
0: Primer Second Edition, W3C Recommendation 28
October 2004. URL: http://www.w3.org/TR/
xmlschema-0/
Gançarski, A., Doucet, A., Henriques, P., 2006. AG-based
interactive system to retrieve information from XML
documents, IEE Proceedings Software Journal,
Volume 153, Issue 2, p. 51-60, April 2006.
Gançarski, A., Henriques, P., 2006. A Formal Definition
of Selection Operations that Extend XQuery with
Interactive Query Construction. International
Conference in Web Information Systems and
Technologies 2006 (Webist06), Setubal, Portugal,
INSTICC Press.
Gançarski, A., Henriques, P., 2007. Using data together
with metadata to improve XML information access.
International Conference in Web Information Systems
and Technologies 2007 (Webist07), Barcelone, Spain,
INSTICC Press.
Manola, F. and Miller, E., 2004. RDF Primer W3C
Recommendation 10 February 2004. URL:
http://www.w3.org/TR/rdf-primer/.
Prud’hommeaux, E. and Seaborn, A., 2007. SPARQL
Query Language for RDF W3C Proposed
Recommendation 12 November 2007. URL:
http://www.w3.org/TR/rdf-sparql-query/.
ITERATIVE XML SEARCH BASED ON DATA AND ASSOCIATED SEMANTICS
483

Figure 1: MEC Ontology.
ICEIS 2008 - International Conference on Enterprise Information Systems
484