We demonstrate the interactive ViziQuer environment for composing rich visual queries (including aggregation, nesting and data expressions) over SPARQL endpoints. We describe the interactive means for building the visual constructs of the query language and discuss the user study results on the query environment usability. The ViziQuer tool environment is publicly available and open source.
Visual query composition paradigm (cf. [
The recent ViziQuer notation ([
We shall report in this paper and present in the demonstration:
Interactive rich visual query building environment with class tree basis and
context-based query construction options;
Context-dependent code completion of attribute and condition expressions and
property paths;
Ready-to-use data schemas over a number of existing public Linked data
endpoints (the query examples in this paper are over Scholarly data endpoint [
The user studies identifying the strengths as well as remaining areas of necessary attention within the visual query creation environment are discussed, as well.
There are interactive multi-modal visual query creation environments available in
Optique VQs [
The open source ViziQuer query environment and its supporting resources are available from the home page http://viziquer.lumii.lv/, including the tool local set up options.
The visual queries are formulated in the context of a data dictionary that is a part of the data schema used by the ViziQuer tool (cf. Section 3), the data dictionary maps the entity short names used in the visual query presentations to their full IRIs.
Each query is a connected graph with one main query node (orange round rectangle).
The query examples in Table 1, formulated over Scholarly Data schema [
The visual links from the union node [+] describe the data links from the Conference instances. 7. Find the keywords used in proceeding papers of both iswc and eswc conferences in 2017, with respective usage counts, order descending by the usage count in iswc 2017.
The unit node [] is a wrapper around the two subquery results (there can be combining, filtering, or even further aggregation within wrapper nodes). 3
The query environment serves the creation of queries (such as the ones listed in Table 1) over a knowledge graph schema describing the structure of the data to be analyzed.
The query environment setup requires loading the data schema, listing the available entity names and connections. The data schema can be extracted either from an OWL ontology or directly from the SPARQL endpoint. There is a list of pre-built schemas, as well as the reference to the schema extraction service available at the ViziQuer Schema Store1. The Scholarly data schema used in the Section 2 examples has been retrieved into the Schema Store from the Scholarly Data SPARQL endpoint2. There is also an example project with the Scholarly data queries available in the Schema Store.
The reference to a SPARQL endpoint is to be specified within a project (in the project parameter window), if the queries are to be directly executed over the endpoint.
The query creation process typically starts by double clicking a node in the class tree (Fig. 1, left) thus adding a query node (orange round rectangle) with this class into the query pane. Figure 1, center, shows the context menu available for nodes within the visual diagram. The attributes, ascribed to the class (or to its superclass, or a subclass) in the data schema, can be checked to be added to the query output via the Add Attributes context menu item (the attribute choice dialogue example is in Figure 1, right; the (select this) option adds the class instance URI itself; the [+] button expands the choice list to offer object property selection options, as well).
At this point the query SPARQL form can be generated or the query can be executed over the provided SPARQL endpoint using the respective context menu commands. 1 http://viziquer.lumii.lv/schema-store/index.html 2 http://www.scholarlydata.org/sparql/
There are also separate context menu items for aggregation (e.g. count) option and condition introduction into the query nodes. The full editing facilities of the visual query element properties are available in the designated property pane that can be used for structural element introduction, fine-tuning, editing and deleting alike.
There are two options of linked node introduction into a query: either by introducing both nodes (typically, classes) into the query pane and then connecting the nodes by a link from the diagram symbol palette, or the Add Link option from the class context menu that suggests adding the links with schema-defined properties for the class, together with the property target classes. Both link creation options distinguish the class join links from the nested query links (links with black bullets, as e.g. in Table 1 examples 4, 5, 6 and 7). The Add Link dialogue further on offers a wizard for fast introducing of aggregation into the nested query, together with its result handling within the host query, as e.g. in query 4 in Table 1.
The textual expression information entry within attribute expression and condition fields and in line property paths is supported by a code completion option that is aware of the expression focus placement with respect to the data schema and the field position within the query. So, entering the property paths e.g. in queries 3, 5 and 6 in Table 1, each property name in the path is gradually offered to the end user after the entry of the previous property name, followed by the dot that indicates the continued navigation.
The query re-shape options involve a service for changing the query main class, and a more fundamental Add Outer Query option that adds a non-data node ‘[ ]’ introducing the outer query level around the current base query, as e.g. in query 7 in Table 1. 4
An earlier user study with Computer Science master’s degree students has shown [
The ViziQuer tool is open source. To ease the local ViziQuer server usage, a prebuilt Docker environment with the tool is offered. The further usability improvements including the integrated and more efficient data schema retrieval is work in progress.