=Paper=
{{Paper
|id=None
|storemode=property
|title=RExplorator - supporting reusable explorations of Semantic Web Linked Data.
|pdfUrl=https://ceur-ws.org/Vol-658/paper471.pdf
|volume=Vol-658
|dblpUrl=https://dblp.org/rec/conf/semweb/CohenS10
}}
==RExplorator - supporting reusable explorations of Semantic Web Linked Data.==
https://ceur-ws.org/Vol-658/paper471.pdf
RExplorator - supporting reusable explorations of
Semantic Web Linked Data.
Marcelo Cohen, Daniel Schwabe
Pontifical Catholic University of Rio de Janeiro
R. M. S. Vicente 225
Gávea, Rio de Janeiro, RJ, Brazil
+55 21 3527-1500
{mcohen21@gmail.com, dschwabe@inf.puc-rio.br}
Abstract. This demo presents RExplorator, an environment which allows non-
technically savvy users, but who understand the problem domain, to explore
the data until they understand its structure They employ a combination of
search, query and faceted navigation in a direct manipulation, query-by-
example style interface. In this process, users can reuse previously found
solutions by other users, which may accomplish sub-tasks of the problem at
hand. It is also possible to create an end-user friendly interface to allow them
to access the information. Once a solution has been found, it can be
generalized, and optionally made available for reuse by other users. This
enables the establishment of a social network of users that share solutions for
problems in particular domains (repositories) of interest.
Keywords. RDF, exploratory search, exploration, ontology, semantic web,
reuse, interface, set-based navigation
1 Introduction
The availability of Linked Open Data in the WWW has increased tremendously1.
Currently, when building a new application, it is becoming increasingly common to
first explore available data that can be leveraged to enhance and complete one’s own
data to provide the desired functionality. The BBC Music website2 is one visible
example of this approach, combining MusicBrainz and DBPedia with their own data.
In previous work we developed Explorator [1], a model for representing
information processing by users in exploratory tasks, and its associated tool, which
1
http://linkeddata.org
2
http://www.bbc.co.uk/music
provides a browser interface supporting this model. Explorator is based on the
metaphor of direct manipulation of information in the interface, with immediate
feedback of user actions. In this demo we present RExplorator 3 , a significant
extension of Explorator, with new features illustrated in this demo.
2 REXPLORATOR
Consider two simple tasks to be carried out over the “Dogfood” data server4,
containing collected publication information for several conferences related to the
Semantic Web - Finding all publications of a given author and finding co-workers of
a given researcher, and their publications We assume the user has no prior
knowledge about the contents of this repository.
For the first task, the user has to
1. Find a class that represents persons
2. Find the desired person, “a”.
3. Find a property “p” that relates a person to publications,
4. Find all triples of the form and collect all objects from these
triples.
In Explorator, this is achieved by first clicking on “Menu”-> All RDF Classes”,
noticing class Person, mousing over it to click on “All Instances”, which reveals a set
of all Persons. Double-clicking on a Person (e.g. “Steffen Staab”), a new box appears
with all details for this resource (i.e., all triples with this resource as subject).
Looking at the details, one notices the property “made”, which relates Person to
Publications. To get all publications by a Person, one may click on the “Selected
Person Details” box, and click on the “S” operand position at the top; click on the
“made” box and click on the “P” operand position at the top, and then clicking on the
“=” (“compute”) operator at the top.
For the second task, the user has to follow a similar set of steps, re-using task 1
once the co-workers have been found.
RExplorator extends Explorator by (we don’t detail all for space reasons)
1. Allowing operations to be parameterized;
2. Allowing the results of a query to be fed as input of another query, thus
forming graphs of interconnected operations;
3. Allowing keeping such graphs as separate workbenches, while enabling
interconnection of graphs across workbenches;
3
Available at http://www.tecweb.inf.puc-rio.br/rexplorator, which includes a demo
video. See also a demo of Explorator - http://blip.tv/play/ AennPpPBIw
4
http://data.semanticweb.org
4. Allowing the designer to import previously defined query graphs into the
current workbench;
5. Allowing the designer to define additional operators beyond the set
operations provided;
6. Allowing the designer to define interfaces oriented towards end users,
hiding details and customizing the look-and-feel.
The original Explorator metaphor lets users compose operations incrementally,
seeing the results at each composition step. Each new query takes its operands from
existing query results. In the end, one may regard this set of inter-related operations
as a graph, similar to an Excel spreadsheet. However, the operations are all grounded,
which would be akin to not having any variables in the formulas of the analogous
spreadsheet. Thus, the first generalization made was to allow operations to have its
operands parameterized, and to propagate values trough the graph of operations when
the value of the parameter is changed. This is equivalent to introducing variables in
the expression that denotes the operation.
Consider step 4 in task 1, finding all publications of a Person. In Explorator, this is
achieved by selecting an instance of Person (e.g., “Steffen Staab in box “All
Persons”) in Figure 1, setting it as the subject parameter, selecting the relation
“make” as the property parameter, and clicking on the “=” operator to find all triples
of the form < made? o> clicking on the icon in each box,
as shown in Figure 1 reveals the actual operations and their dependencies .
Figure 1 – Query structure and dependencies
The first box, Selected Person Details, represents the query that finds out all
triples with a given Person as subject. Notice that the first position, “S”, has been
parameterized, and the current parameter value is (the URI for) Steffen Staab. If we
drag any person from the rightmost box (All Persons) onto the “S” position in the
Selected Person Details box, the value is replaced and the query re-evaluated.
The Publication by Person query (middle box) is defined as taking its “subject”
parameter from the “subject” position of the Selected Person Details query.
Therefore, if a new value is plugged into the “S” position in the Selected Person
Details query, it is automatically propagated to this query, which triggers its
reevaluation.
RExplorator organizes the workspace into workbenches, each representing a task.
A user may save workbenches for later reuse, and share it with other users as well.
The development interface of RExplorator is best suited to allow users to explore
RDF repositories, and requires understanding the RDF model. RExplorator allows
expert users to provide an end-user friendly interface – called the Application
Interface - to solutions found while exploring datasets. This interface is generated by
a combination of views, which are defined using the “views” menu option; generic,
pre-defined views are initially available for reuse. Views make full use of CSS,
which is also defined in a separate view that can be customized to change the look-
and-feel of the generated interface. Figure 2 shows an example of an application
interface.
Figure 2 – User-friendly interface with All Persons and Selected Person Co-Workers.
ACKNOWLEDGMENT. Daniel Schwabe was partially supported by a grant from CNPq.
3 REFERENCES
[1] Araújo F. C. S.; Schwabe D.; Explorator: A tool for exploring RDF data through direct
manipulation, In: Proceedings of the Linked Data on the Web Workshop (LDOW2009),
Madrid, Spain, April 20, 2009, CEUR Workshop Proceedings, ISSN 1613-0073, online
http://CEUR-WS.org/Vol-538/ldow2009_paper2.pdf