Health care and life sciences research heavily relies on the ability to search, discover, formulate and correlate data from distinct sources. Although the Semantic Web and Linked Data technologies help in dealing with data integration problem, there remains a barrier adopting these for non-technical research audiences. In this paper we present FedViz, a visual interface for SPARQL query formulation and execution. FedViz facilitates both federated and non-federated SPARQL queries formulation through a visual interface and uses FedX for query execution and the results retrieval. We evaluate the usability of our system by using the Standard System Usability scale. Overall usability score of 74.16 suggests that FedViz interface is easy to learn, consistent, and adequate for frequent use.
FedViz allows the user to select concepts and properties from multiple datasets using nodes and edges and makes SPARQL query in a background. Furthermore, it also allow users to edit the resultant SPARQL query before sending it to the the query execution engine. The assembled SPARQL query is then executed through FedX [12], a well-known SPARQL query federation engine. Finally, the query execution results are displayed by FedViz. Note FedX can executes both federated (without using SPARQL SERVICE clause) and non-federated SPARQL queries. Current version of FedViz supports six real time Life Science data sources namely Kegg (http://www.genome.jp/kegg/), Drugbank (http://www.drugbank.ca/), DailyMed (http://dailymed.nlm.nih.gov/dailymed/ index.cfm), Medicare (http://wifo5-03.informatik.uni-mannheim.de/medicare/), Sider (http://wifo5-03.informatik.uni-mannheim.de/sider/), and Diseasome (http://wifo5-03. informatik.uni-mannheim.de/diseasome/). 2
FedViz (Demo available at: http://srvgal86.deri.ie/FedViz/index.html) is a browser based application currently supported by IE, Firefox and Chrome, that provide users with a flexible visual interface to formulate and execute SPARQL queries. FedViz visualises six life sciences datasets and facilitate users to draw meaningful biological correlations including drug-drug interaction, drug-disease interaction, drug-side effect interaction and disease-gene interaction. FedViz supports full range of SPARQL1.1 constructs either selected through interface or can be directly added to the query in SPARQL Editor. 2.1
Query Formulation We demonstrate our visual query formulation and execution process by using an example scenario, i.e., find the drug-disease interaction from the aforementioned data sources. Scenario: Drug-Disease Interaction: Drugs with their inchi key, molecular weight average and chemical structure for curing disease Alzheimer:
This query requires data integration from Drugbank (containing drugs information) and Diseasome (containing disease information) and be formulated by using the following step-by-step approach (ref., Fig. 1): 1. The first step is to identify how Drugbank and Diseasome datasets are connected to each other? This connectivity (i.e., via classes drugbank:drug and diseasome :disease) can be found by using the Class visualisation View of FedViz. 2. User selects Drugbank from the Datasets Selection box (window A). 3. The visualisation for Drugbank dataset can be seen in Window B where user selects drugbank:drug class and its properties (i.e., inchi key, molecular weight average and possible disease target). 4. Step 2 and 3 are now followed for Diseasome dataset, i.e., select diseasome:disease class and it’s name property (window C). 5. Next, FedViz SPARQL Query Editor allows user to add constraints to the formulated query such as select projection variables, apply SPARQL LIMIT, FILTER, ORDERY BY clauses, and can further edit the query according to his choice (window D(a) and D(b)). 6. The final query can be seen on submission (window E).
A: Datasets selection
B: Drugbank Concept & Property selection D(a):Query editor
D(b):Query editor
E:Formulated Query F:Federated query results
G:Instance data exploration 7. Query is executed over FedX and the retrieve results are displayed by FedViz (Result window F). 8. Finally, by selecting any URI from the retrieved result, FedViz provided detailed information regarding that entity (window G).
To assess the usability of FedViz, we used the standardised, System Usability Scale (SUS) [10] questionnaire2. The SUS is a low-cost, reliable usability scale that can be used for global assessments of systems usability[10]. The survey was filled by 15 users3 including researchers and engineers in Semantic Web (by 10th July 2015). We achieved a mean usability score of 74.16 indicating a high level of usability according to the SUS score. The average scores (out of 5) for each question along with standard deviation is shown in Figure 2. The responses to question 1 suggests that FedViz is adequate for frequent use (average score to question 1 = 3.8 0.86) by users all of type. The responses to question 7 (average score 4.06 0.96) suggests that most people would learn to use this system very quickly and the responses to question 3 indicates that FedViz is easy to use (average score 4 0.84). 2 Our survey can found at: http://goo.gl/forms/bhReuNgd6O 3 Users from AKSW, University of Leipzig and INSIGHT Centre, National University of Ireland, Galway. Summary of the responses can be found at: https://goo.gl/ZOrJx9
I needed to learn a lot of things before I could get going with this system (
I felt very confident using the system (
I found the system very cumbersome to use (
I thought there was too much inconsistency in this system (
I found the various functions in this system were well integrated (
I thought the system was easy to use (
I found the system unnecessarily complex (
The work presented in this DEMO paper has been partly funded by Science Foundation Ireland under Grant No. SFI/08/CE/I1380 (Lion-2).