USING DATA TOGETHER WITH METADATA TO IMPROVE

XML INFORMATION ACCESS

Alda Lopes Gançarski

GET/INT, CNRS Samovar

9 rue Charles Fourier, 91011 Évry, France

Pedro Rangel Henriques

University of Minho

Department of Informatics, 4710 Braga, Portugal

Keywords: XML, XQuery, ontology, Dublin Core, metadata, SPARQL.

Abstract: In this paper, we address the use of data together with metadata to improve information access to XML

document collections. We first emphasize the possibility to associate meta-information to parts of XML

documents, and not only entire documents. This is accordingly to the fact that XML elements are considered

as retrieval units. We, then, propose to explicitly search the desired information using a query language that

is composed of XML search and metadata search. We focus our ideas in two types of metadata: Dublin Core

and ontologies.

1 INTRODUCTION

XML information access is done using structured

query languages such as XPath (Berglund et al.,

2005) and XQuery (Boag et al., 2006), the standard

proposed by the W3C. These query languages are

being extended with the possibility of associating a

score to an expression that verifies if some phrase

exists in the content of some element, as in

traditional information retrieval. This functionality is

included in the Full-Text language also proposed by

the W3C (Amer-Yahia et al., 2006).

To improve data processing, document

collections and Web resources are associated with

semantic descriptions, or metadata. In order to be

able to exchange the semantics of information, one

first needs to agree on how to explicitly model it.

This can be done using a standard set of

characteristics, like title or author, or using a more

sophisticated description in the form of ontologies.

In the first case, a standard set of such characteristics

was developed in the Dublin Core Metadata

Initiative (Dublin Core Metadata Initiative, 2006).

This set is composed, among other concepts, of

Creator, Date, Format, Language, Publisher, Title,

Subject and Keywords, Abstract.

Concerning the second case, 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. To

formally define ontologies, W3C has proposed the

Resource Description Language (RDF) (Manola

and Miller, 2004), which allows the representation

of metadata about Web resources.

Metadata search can be done using a simple

natural language expression (Finin et al., 2005),

using a navigational structure (Fluit et al., 2005) or

using a dedicated query language (Corby et al.,

2004) like SPARQL (Prud’hommeaux and Seaborn,

2006), the W3C recommendation for the standard

query language for RDF. In general, works are

devoted to search information directly in documents

or indirectly in ontology’s concepts to get associated

resources, not using data and metadata together in

the search process. In this article we exploit the use

328

Lopes Gançarski A. and Rangel Henriques P. (2007).

USING DATA TOGETHER WITH METADATA TO IMPROVE XML INFORMATION ACCESS.

In Proceedings of the Third International Conference on Web Information Systems and Technologies - Internet Technology, pages 328-332

DOI: 10.5220/0001262003280332

 SciTePress

of XML documents together with the respective

metadata to access information. We believe that

both may be interesting to the user and can help him

to find interesting information.

Metadata descriptions may be done at elements

level, giving semantic information for elements

retrieval. This is discussed in Section 2. Section 3

explains our proposal for the search based on both

data and metadata showing some examples. The

article finishes with a brief conclusion, indicating

some future work.

2 ELEMENTS METADATA

In accordance with XQuery, XML elements are

retrieval units and, thus, it is interesting to allow for

metadata descriptions associated to elements, instead

of only having document metadata.

To exemplify elements metadata, suppose a book

where chapters, not only are written by different

authors, but also cover many sub-subjects of the

book’s main subject. So, different meta-information

can be associated to each chapter. Also, a useful

metadata concept can be the date of the last update

of an element.

If RDF is used to specify Dublin Core (DC)

metadata and ontologies, elements can be referenced

using URI references. A URI reference (or URIref)

is a URI together with an optional fragment

identifier at the end. For example, the URI reference

http://www.example.org/article.xml#section2

corresponds to the second section element in the

article.xml document.

3 XML COMPOSED SEARCH

To perform information search based on both data

and metadata, some extensions to XQuery must be

done. The next sections explain such extensions

when there is DC metadata and when metadata is in

the form of ontologies, respectively.

3.1 Using Dublin Core Metadata

If metadata follows the DC proposal, metadata can

be expressed in different ways.

When documents are in HTML format, DC

metadata can be embedded in the document using

the special Meta tag. An example of a Meta tag is in

the following document:

<Html>

<Head>

<Title>XML standard</Title>

<Meta Name=”DC.Creator”

Content=”Paul”>

</Head>

<Body>

<H1>XML Applications</H1>

<P>XML stands for eXtensible …</P>

</Body>

</Html>

In this example, metadata indicates that the creator

of the page is “Paul”. This is indicated in attributes

Name and Content of the Meta tag, respectively.

Restrictions on metadata are done accessing Meta

tags’ attributes. Thus, no extensions to XQuery are

needed, as indicated in Figure 1.

Figure 1: Using XQuery to query HTML documents

including the DC <Meta> tag.

For example, to get HTML documents created by

Pearl, the XQuery query is:

for $a in doc(“http://...”)

/Html[.//Meta/@Name=”DC.Creator” and

.//Meta@Content=“Pearl”]

return $a

When DC metadata is expressed in RDF, we

propose an extension to XQuery in order to allow for

the inclusion of metadata restrictions in structural

restrictions. These restrictions are expressed using

the so called metadata: functions.

As depicted in Figure 2, having a processor for

XQuery extended with the metadata: functions, the

user can make queries over XML collections

associated to DC metadata expressed in RDF.

quer

resul

XQuery

Processor

HTML +

<Meta>

USING DATA TOGETHER WITH METADATA TO IMPROVE XML INFORMATION ACCESS

329

. 3

Figure 2: Using XQuery extended with metadata:

functions to query XML documents associated to DC

metadata expressed in RDF.

For example, suppose the following XML

document:

<Book>

<Title>XML Guide</Title>

</Chapter>

</Chapter>

…

</Book>

Suppose this document has the following DC

metadata expressed in the form of RDF triples (for

simplicity, we ignore prefixes that correspond to

URIs):

Book.xml#Chapter1 Creator “Kevin”.

Book.xml#Chapter2 Creator “Loik”.

To get chapters written by Kevin, a simple XQuery

query could be:

for $c in doc(“http://...”)

/book/chapter[metadata:Creator(.)=“Kevin”]

return $c

In this query, a filter is imposed to chapters. In this

filter, function metadata:Creator() returns the value

of the Creator concept in the metadata description of

a chapter. The chapter is passed to the function using

the “.” symbol, which expresses the current element

being evaluated. Note that metadata: prefix indicates

that Creator is a function about metadata.

Suppose now that there is information about the

subjects treated in the chapters:

Book.xml#Chapter1 Subject “SGML”.

Book.xml#Chapter2 Subject “HTML”.

If the user wants chapters about SGML, he can make

the following XQuery query:

1 for $c in doc(“http://...”) /book/chapter

2 score $s as

3 $c ftcontains “SGML” OR

4 metadata:Subject($c) ftcontains “SGML”

5 order by $s

6 return $c

In this query, the for clause stores in variable $c all

the book chapters (line 1). Then, the score clause

associates to each chapter a score stored in variable

$s (line 2). The score represents how much the

chapter is about the desired subject “SGML”. In

XQuery, the ftcontains expression verifies if some

phrase exists in an element. If it is included in a

score clause, it verifies how much the element is

about the subject expressed in the phrase, i.e., the

score of the element with respect to the subject. In

the example query, the computation of the score

takes into account both the content of the chapter

(line 3) and the Subject concept of its metadata

description (line 4). This can help making better

score computations. Finally, the result of this query

is the list of chapters (line 6) ordered by their score

(line 5). In general, for each DC concept, a function

is used to access to the respective metadata value.

XQuery includes function calls. Functions can be

XQuery pre-defined ones or user defined functions,

such metadata ones.

Each DC concept is associated to a mapping, or

table, from elements (or nodes) to metadata values.

This table is created when DC metadata is

associated to elements. Suppose the mapping of

concept X is:

MapX : node() × xs:string

The XQuery node test node() matches any node. A

value is generaly represented by a string which is

denoted here by xs:string. The metadata function

metadata:X can, then, be defined by:

declare function metadata:X ($a as node())

as xs:string

{

return MapX[$a]

}

quer

resul

XQuery +

metadata:

functions

Processor

XML

DC RDF

WEBIST 2007 - International Conference on Web Information Systems and Technologies

330

Here, $a is the variable that stores the node. The

result of the function is the value corresponding to

the node in the MapX table.

3.2 Using Ontologies

When a XML document is associated to semantic

descriptions in the form of ontologies expressed in

RDF, we propose to integrate SPARQL queries in

XQuery queries. This can be done by adding a new

clause metadata to the for clause of both languages.

As depicted in Figure 3, the user can make queries

over XML collections associated to RDF metadata

using a processor for XQuery extended with the

metadata clause for SPARQL queries.

Figure 3: Using XQuery extended with metadata clause to

query XML documents associated to RDF metadata.

Suppose the following ontology including references

to elements of an XML document.

Book.xml about Animals.

Book.xml#Chapter1 about Fishes.

Book.xml#Chapter2 about Birds.

Book.xml#Chapter3 about Mammals.

Book.xml#Section1 about Men.

Fishes is Animals.

Birds is Animals.

Mammals is Animals.

Men is Mammals.

Men eats Birds.

men eats Fishes.

If the user wants sections about animals that are

eaten by men, he can specify the following XQuery

query:

1 for $s in

2 doc(“http://.../Book.xml”) /book//section

3 metadata $s in

4 SELECT ?s WHERE (?o eats Fiches)

5 (?o eats Birds)

6 (?s about ?o)

7 return $s

In this query, the for clause associates to variable

$s the set of sections of the document (line 1 and 2).

The metadata clause (line 3) includes a SPARQL

query (line 4 to 6). This SPARQL query selects all

the elements (stored in variable ?s) that are about

some subject (stored in variable ?o) which eats

fishes or birds. The resulting set of elements of the

internal SPARQL query (in variable ?s) is

intersected with the set of elements of the XQuery

external query (in variable $s) to get the desired

sections of the book.

To show another example, suppose the following

products catalogue:

<product><name>Water Corola</name>

</product >

<product>… </product> …

</products>

Suppose also that this document is associated to the

following RDF description:

Catalog.xml#product1 lastUpdate “10/1/2005”.

Catalog.xml#product2 lastUpdate “10/2/1990”.

To get products which last update is before 1990, the

query is:

1 for $p in doc(“http://.../catalog.xml”) //product

2 metadata $p in

3 SELECT ?p

4 WHERE (?p lastUpdate ?u)

5 AND ?u < “1/1/1990”

6 return $p

In this query, variable $p stores the product elements

of the document (line 1). The SPARQL query (lines

3 and 4) stores in variable ?p elements which last

update, stored in variable ?u, is before 1990. The

content of variable $p in the external XQuery query

is intersected with the content of the

?p variable in

internal SPARQL query, yielding the set of desired

products.

SPARQL queries are included in the XQuery

grammar extending the production that derives the

FLWOR expressions with the metadata clause. The

current production is:

[33] FLWORExpr ::= (ForClause |

LetClause)+ WhereClause?

OrderByClause?

“return" ExprSingle

XML

RDF

quer

resul

XQuery +

metadata

clause

Processor

USING DATA TOGETHER WITH METADATA TO IMPROVE XML INFORMATION ACCESS

331

. 5

Extending it with metadata restrictions yields:

[33] FLWORExpr ::= (ForClause |

LetClause)+ MetadataClause?

WhereClause? OrderByClause?

"return" ExprSingle

Symbol MetadataClause derives the metadata

clause by the following productions:

MetadataClause ::= “metadata”

“$” VarName (“,” “$” VarName)*

“in” SPARQLQuery

The symbol VarName belongs to the XQuery

grammar and derives a variable name. In its turn,

symbol SPARQLQuery derives a SPARQL query

using the grammar defined in (Prud’hommeaux and

Seaborn, 2006).

4 CONCLUSION

In this paper, we informally propose to integrate data

and metadata search in the same query language. We

believe this can help users to get useful information.

As future work, we intend to formalize the proposed

XPath and XQuery extensions, like the metadata

clause. This formalization will take into account

many aspects concerning metadata. For example,

intra- and inter-document links must be considered.

Once the extension is formalized, we intend to

create a prototype processing environment for the

extended XQuery. We can use existing XQuery and

SPARQL query processors integrated with editing

and results visualization. We intend, then, to test the

prototype and the usefulness of the approach using

existing document collections and respective

metadata.

REFERENCES

Amer-Yahia, S., Botev, C., buxton, S., Case, P., Doerre, J.,

Holstege, M., McBeath, D., Rys, M.,

Shanmusgasundaram, J., 2006. XQuery 1.0 and XPath

2.0 Full-Text W3C Working Draft Draft 1 May 2006,

http://www.w3.org/TR/xquery-full-text/.

Berglund, A., Boag, S., Chamberlin, D., Fernandez, M.,

Kay, M., Robie, J., Siméon, J., 2006. XML Path

Language (XPath) 2.0 W3C Proposed

Recommendation 21 November 2006,

http://www.w3c.org/TR/2006/PR-xpath20-20061121.

Boag, S., Chamberlin, D., Fernandez, M., Florescu, D.,

Robie, J., Siméon, J., 2006. XQuery 1.0: An XML

Query Language W3C Proposed Recommendation 21

November 2006, http://www.w3c.org/TR/2006/PR-

xquery-20061121/.

Corby, O., Dieng-Kuntz, R. and Faron-Zucker, C., 2004.

Querying the Semantic Web with Corese Search

Engine, 3rd Prestigious Applications Intelligent

Systems Conference (PAIS), Valencia, Spain, 2004.

Dublin Core Metadata Initiative, 2006. URL:

http://dublincore.org/ (last updated: 18 December

2006).

Finin, T., Ding, L., Pan, R., Joshi, A., Kolari, P., Java, A.

and Peng. Y., 2005. Swoogle: Searching for

knowledge on the Semantic Web, National

Conference on Artificial Intelligence (AAAI) 2005,

Intelligent Systems demo, Pittsburgh, Pennsylvania,

USA, July 2005.

Fluit, C., Sabou, M. and Harmelen, F., 2005. Ontology-

based information visualization: towards semantic web

applications. In Visualising the semantic web (2nd

edition), Valdimir Geroimenks Editor, Springer

Verlag, 2005.

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., 2006. SPARQL

Query Language for RDF W3C Working Draft, 4

October 2006. URL: http://www.w3.org/TR/rdf-

sparql-query/.

WEBIST 2007 - International Conference on Web Information Systems and Technologies

332