Calcolus algorithms for the ARTEMIS module

Here is an analysis oriented to the software implementation of the steps performed by ARTEMIS during calcolus:

Phase Name	Input	comments	Output
Initial Distance calcouls	Thesaurus relations Relations weights	Transforms the list of ThesRelation in a Map of distance between objects using the according to the Relations weights.	Initial Distance Matrix This is a matrix of n per n elements where n is the cardinality of the set union of all the Attributes and all the interfaces of the full schema
Name Affinity calcouls	Shema description (list of all interfaces) Initial Distance Matrix	This is a backtracking algorithm that, starting from the intitial distance matrix computes lightest path between all nodes of the graph. Nodes of the graph are all the Attributes and all the interfaces of the full schema	Name Affinity Matrix This is a matrix of n per n elements where n is the cardinality of the set union of all the Attributes and all the interfaces of the full schema
Structural Affinity calcouls	Shema description Name Affinity Matrix Thesaurus relations Affinity threshold	Sum an multiplication formulae that for each pair of Interfaces in the shema computes a value based on input data.	Structural Affinity Matrix
Global Affinity calcouls	Structural Affinity Matrix Name Affinity Matrix Na, Sa coefficients	Sum an multiplication formulae that for each pair of Interfaces in the shema computes a value based on input data.	Global Affinity Matrix
Clustering	Cluster threshold Global Affinity Matrix	Sum an multiplication formulae that for each pair of Interfaces in the shema computes a value based on input data.	Clusters Cluster tree ? raw MappingTable ?

Implementation steps

Things done

Implementation of the classes Distance, DistenceMap and DistanceMapSymmetrical to replace matrices with indexed data structures (for more flexibility).

Things to be done

Name Affinity calcouls
Structural Affinity calcouls
Global Affinity calcouls
Clulstering calcouls

The MOMIS Home Page