Database Reference
In-Depth Information
JSP, etc. Mashups are different from traditional web applications because they are usu-
ally dynamically created to serve a very specific and short lived task. Several mashup
editors have been launched to encourage people to build new applications using the
massive amount of publicly available contents. Yahoo Pipes
4
, Google Mashups
5
and
IBM Mashup Center
6
are a few examples of the popular mashup editors. However, the
limitation of existing mashup editors is that they focus only on web feeds or API's.
These web feeds can represent simple information but lack the capability to represent
or query data items provided by querying interfaces or data services [2]. On the other
hand, API's are usually limited to a specific application thus requiring different imple-
mentations for each of the sources used in the mashups. Currently, the development of
data mashups to deal with complex data structures requires strong programming skills,
making mashups hard to create for novice users.
We utilize the concept of data mashups and use it to dynamically integrate hetero-
geneous web data sources by using the extension of XQuery proposed in [3]. All the
available data sources over the internet are considered as a huge database and each data
source is considered as a table. Data mashups can generate queries in extended XQuery
syntax and can execute the sub-queries on any available data source contributing to
the mashup. XML and RDF are the prevailing data formats for web data sources. To
query these data sources, one can use XQuery and SPARQL - their respective query
languages. The novelty of our tool is that it integrates the powerful features of database
querying into a data mashup tool. It provides an easy to use interface of a mashup edi-
tor to generate complex queries visually for the integration of a multitude of distributed,
autonomous, and heterogeneous data sources.
2
Database Oriented Mashups
A mashup application comprises three major components, which are (1) data level, (2)
process level, and (3) presentation level [4]. The data level is mainly concerned with
accessing and integrating heterogeneous web data sources. These sources can provide
structured, semi-structured or unstructured data. Existing data mashup tools cannot deal
with structural and semantic diversities of heterogeneous data sources. Recently, the
importance of using data mashups for data integration using database oriented mashups
has been realized [2]. Inspired by Yahoo pipes, there are a few attempts such as MashQL
[5] and Deri Pipes [6] to generate semantic queries from data mashups. However, to
the best of our knowledge, there exists no data mashup tool which allows the user to
formulate queries over web data sources using their respective query languages and at
the same time deals with the heterogeneity of the data sources. Our tool is similar to
MashQL and Deri Pipes, but we focus on the XQuery extension of [3] with additional
support of the SPARQL query language. Using our approach, existing data integration
support for mashups is further enhanced to formulate a single query containing inside
sub-queries of different query languages to deal with heterogeneous data integration.
4
http://pipes.yahoo.com/pipes
5
http://code.google/com/gme
6
www.ibm.com/software/info/mashup-center/
