FLuID: A Meta Model to Flexibly Define Schema-level Indices for the Web of Data
share
Schema-level indices are vital for summarizing large collections of graph data. There is a large variety in how existing schema-level indices capture the schema of data instances. Each index has its value for a particular application scenario or information need. However, existing indices define only a single, fixed data structure that is tailored to a specific application scenario. Thus, the indices cannot be easily adapted or extended to changing requirements. In order to address these shortcomings, we propose a formal, parameterized meta model called FLuID (Flexible schema-Level Index model for the web of Data) that allows to quickly define, adapt, and compare different schema-level indices for distributed graph data. We conduct an extensive study of related works and abstract from the features of the existing index models to FLuID. In addition, FLuID provides novel features such as aggregation of instances over owl:sameAs. We conduct a detailed complexity analysis and show that indices defined with FLuID can be efficiently computed on average in Θ(n) w.r.t. n being the number of triples in the input data graph. Furthermore, we implemented the FLuID meta model following an existing stream-based schema computation approach for the Web of Data. We empirically analyze different index models for different application scenarios, types of queries, datasets, and space requirements. This provides for the first time in-depth insights for understanding the influence of design choices of the index models and their usefulness in different scenarios.
READ FULL TEXT