SPARE-PART is a Scalable Real Time Spatio-Temporal Aggregate Query Processing Framework using Parallel Data Representation Approximation. It is a novel Spark based framework featuring a generic handling of spatio-temporal queries by using multiple representations to answer aggregate queries with real time update of query answer and error estimate.

The query processing utilizes a continuous on-demand conversion of input data into multiple representation streams that are handled by different distributed execution engines.

Individual execution engine applies a correspondingly extendable algorithm to estimate relative error using combined index traversal and sampling. The result aggregate value and error are continuously updated by combining outputs of execution engine on different representations and displayed to user interface.

The representations include vector based, histogram based, interval based and are maintained in-memory by Spark for faster retrieval and future queries.