GraphTQL: A visual query system for graph databases

摘要

End-users who are experts in an application domain need to retrieve data from databases in order to answer particular inquiries. Visual Query Systems (VQSs) facilitate this task; but, formulating complex queries is still challenging for end-users. In particular, VQSs for graph databases can be broadly classified into two groups: One group is based on data exploration and allow for direct manipulation of data graphs, but with limited query expressiveness. The other group, oriented to build visual sentences that represent the query, provide greater expressiveness, but they transfer complex language concepts and operations to the visual language. Thus, this research developed GraphTQL a visual query system that provides greater expressiveness and ease in the query formulation. GraphTQL uses a graph data model to depict the domain of interest at the conceptual level. A set of operators provides the means to transform schema and instance graphs, reducing them progressively to obtain the data of interest. The operators include implicit management of incomplete data; as well as clarification dialogues to guide users in order to express complex filter conditions and to select the paths which connect entities in the graph. The development of GraphTQL followed a user-centered design approach which focused on the medical domain. Thus, potential users' contribution and feedback played a significant role in this work. A comparison with a graphical query interface for SPARQL showed better results for GraphTQL regarding the number of queries correctly formulated by users, the number of errors and the time spent for query formulation.