Exploiting Recurrent Retrieval Needs in Querying Heterogeneous and Dynamic Graph Dataspaces

Querying data sources containing heterogeneous and dynamic information is a complex task: high quality answers need to be produced in a limited response time. Dynamic contexts preclude user involvement in interpreting the request and thus solutions should be devised, relying either on additional knowledge about the current execution (e.g., query context or user profile) or previous executions of the same request [?]. The aim of this discussion paper is to exploit similar requests recurring over time for improving query processing, in terms of result quality and processing time, and to provide a general framework for representing and managing information collected in the executions of (sets of) recurring queries. Our approach relies on the enabling concept of Profiled Graph Query Pattern (PGQP), which represents a set of (previously executed) queries associated with information about their executions. Differently from apparently similar proposals (materialized views, smart caching), the proposed approach approximately matches queries with profiled patterns with the aim of retrieving various kinds of information, related to past executions of similar queries, and improving the processing of the query at hand.