There is a reason why the proverb says “A picture is worth a thousand words”: diagrams or sketches of concepts and of the relationships that exist between them help understand their meaning. In the Semantic Web, the concepts and relationships of the world are modeled through ontologies, but current standards for ontology languages such as OWL 2 don't buy into this idea, recurring to formula-based representations, that work well with machines and not so well with humans. Similarly, the languages used to query ontologies, in particular SPARQL, are not immediately intuitive for people who lack specific training or background, so providing graphical support for query formulation can make the life of the end user easier. In this work we present the SPARQLING system for SPARQL query building based on the GRAPHOL visual language for ontologies. The characterizing feature of SPARQLING is the idea to preserve and take advantage of the native diagrammatic representation of GRAPHOL ontologies, allowing the user to navigate it, and to construct a graph-based representation of the query over it through a simple point-and-click mechanism. The system then automatically transforms the graphical query into the SPARQL syntax.
Ontologies provide a conceptualization of a domain of interest, meant to foster
knowledge sharing and integration among different information systems, and to support
human understanding and communication through a high level of abstraction that is clear
for domain experts. Recent years have seen a sharp increase in the popularity of
ontologies in modern applications, underlined by wide-spread use in biomedicine,
ecommerce, Semantic Web, and Data Integration [
Ontologies can be formalized in different languages, from Description Logics [
Aside from acting as a standardized reference model, ontologies can also be used as
a tool for accessing and managing data. In this scenario, the use of triple stores for
managing (Linked) enterprise data is quickly spreading [
In this work we present the SPARQLING system for SPARQL query building based on the GRAPHOL visual language for ontologies. The characterizing feature of SPARQLING, which differentiates it from these previous works, is the idea to preserve and take advantage of the native diagrammatic representation of GRAPHOL ontologies, al
lowing the user to navigate it, and to construct a graph-based representation of the query over it through a simple point-and-click mechanism. The system then automatically transforms the graphical query into the SPARQL syntax. SPARQLING is an open-source project, available on Github4, and is supported by OBDA Systems5.
In the following sections we first provide some background on the GRAPHOL language and on SPARQL, and then we introduce the SPARQLING system. Finally, we conclude the work with a brief discussion on future developments of the system. 2
In this section, we briefly introduce the GRAPHOL [
GRAPHOL is a completely graphical language for representing OWL 2 ontologies
which has an inherent formal semantics based on DLs and resembles ER diagrams,
but which is in fact able to fully capture OWL 2 without recurring to formulas or
annotations. The basic idea behind GRAPHOL is to represent the ontology as a set of
inclusions, as in DLs or OWL 2, while preserving a graphical representation of it, given
in terms of a directed graph. The nodes of the graph are atomic predicates or
graphical operators, whereas the edges are solid or dashed arrows. The latter denote input to
operators, whereas the former denote inclusions. In GRAPHOL, rectangles are used for
classes (called entities in ER), diamonds are used for object properties (corresponding
to binary relationships in ER), circles are used for data properties, and rounded
rectangles are used for data types (e.g., string and integer). GRAPHOL expressions are
either nodes representing predicates of the ontology (e.g., named classes and
properties), or a combination of these nodes with other nodes that represent operators, shaped
as blank or solid boxes, or hexagons. For instance, labeled blank and solid boxes are
used for constructing object or data property restrictions on the domain and range,
respectively, and solid hexagons represent disjoint unions. Figure 1 depicts a portion of
the Base Registry of Individual, Families, and Cohabitation ontology, developed for the
Italian National Institute of Statistics [
SPARQL is the W3C recommendation query language for RDF. Its structure is based on graph patterns and query evaluation consists in subgraph matching. The basic building block of a SPARQL query is a basic graph pattern (BGP), which is a set of triple patters, i.e., RDF triples that may contain query variables at the subject, predicate, and object position. More complex SPARQL query patterns are built over BGPs. 3
SPARQLING is a web-based application for constructing SPARQL queries by
exploiting the GRAPHOL representation of the ontology. The user interface is split in three
different sections, as shown in Figure 2. On the left, the user can see the GRAPHOL
representation of the ontology, while on the bottom right the query canvas displays the
visual representation of the query under construction, and on the top right the system
provides the SPARQL translation of the query. The current implementation of
SPARQLING is tailored towards queries for Ontology-based Data Management systems [
The distinguishing feature of SPARQLING is that the system keeps the GRAPHOL representation of the ontology always accessible to the user. The intuition is that having a clear understanding of the structure and content of the ontology, in its GRAPHOL form, is helpful in designing queries. Indeed, the mental process of writing a query is natural when looking at a GRAPHOL diagram because it recalls the act of tracing a path on it. SPARQLING attempts to capture these query paths on the diagram through the visual representation of the query provided in the query canvas.
This representation, which we refer to as query graph, indeed has the structure of a graph, to naturally recall SPARQL’s BGPs. Generally speaking, a triple pattern in the query corresponds to an edge connecting two nodes, each representing a variable or a constant. To provide a visual correspondence with GRAPHOL, the graphical representation of the ontology predicates in the query graph is the same as in GRAPHOL.
To add a triple to the basic graph pattern of the query, the user double clicks on a predicate symbol in the GRAPHOL diagram. According to the specific predicate symbol, the corresponding edge and nodes are automatically added to the query graph. If a node of the query graph is selected prior to choosing the symbol in the diagram, the new edge will be linked to it. Otherwise, a new disconnected component will be added. Actions on specific nodes of the graph are performed through contextual menus that appear by right clicking on a node, allowing for renaming, deleting, switching between variable and constants, adding to the select clause, and specifying filter expressions.
The system maintains at all times a SPARQL representation of the query which the user can inspect and modify. Each modification of the SPARQL query is immediately mirrored in the query graph, and viceversa. By hovering onto the name of a variable in the SPARQL query, the corresponding node in the query graph is highlighted in order to make its position stand out. The user can actually use SPARQLING by interacting exclusively with the query graph or with the SPARQL representation, according to the user’s preference. The application is developed in JavaScript, using the Cytoscape.js library6 for the rendering of the GRAPHOL diagram and the query graph, and the cola.js library7 for the force-directed layout of the query graph. 4
We have presented the SPARQLING system for drawing SPARQL queries over GRAPHOL ontologies. The development of this system aims to allow users of SPARQLbased query systems to produce queries in a more intuitive fashion. Although the current version is fully functional, the SPARQLING project is still under development and will be extended in different directions. We aim to extend the portion of SPARQL which is currently supported by SPARQLING; we will work to further improve the synergy between the GRAPHOL diagram and the query graph that the user is constructing; and we will allow importing SPARQL queries over the ontology into the system. Finally, testing of the system has so far been carried out through its use in the field; we would like to conduct a more structured and thorough user evaluation to identify potential areas of improvement in the system. Moreover, we plan to conduct a comparative evaluation with other graphical SPARQL editors.