-
Notifications
You must be signed in to change notification settings - Fork 1
Processing Stages
kamir edited this page Jul 4, 2012
·
12 revisions
- Nodes are entities, stored in a database, described by properties in records, eventually spread over different tables or databases.
- Time dependent properties are stored in a time series store (TSS).
- Node data can contain raw data, which is processed by a pre processing stage (PPS), utilizing the cluster
- for a node we store raw data or preprocessed data, together with its pre processing stage parameters (PPSP)
- Links are identified by a link creation mode (LCM), which for itself is parametrized by a set of LCM parameters (LCMP).
- LCM : are program modules, which are deployed to the system as modules or connected as whole subsystems.
- LCMP : is a set of METADATA which also includes one copy of the executable to reproduce the calculation.
- LCM working on just one property per node, a set of properties per node or whole time series of data per node
- the kind of calculation decides what subsystem has to be used (MATLAB, R, CUDA, Hadoop M/R)
- Complex system representation (CSR) are defined by an initial node selection, called NodeGroup (NG)
- a Complex systems view (CSV) is defined by a combination of a CSR and LCM including its special LCMP
- CSV : are described by a set of METADATA, calculated by LCM, and represented by a link set (LS)
- CSV : can be prepared to track a complex systems properties in time
- CSV : are created by implementations of LCM, either by MATLAB programs, CUDA programs or Map Reduce programs on a Hadoop systems using Mahout or Giraph
- CSV : are processed in the same fashion like they are created but with different algorithms
- Complex systems view (CSV) is defined by a combination of a CSR and LCM including its special LCMP
- CSV : are described by a set of METADATA, calculated by LCM, and represented by a link set (LS)
- CSV : can be prepared to track a complex systems properties in time
- CSV : are created either by MATLAB programs, CUDA programs or Map Reduce programs on a Hadoop systems using Mahout or Giraph