[Affiliation Propagation] #162

miriam.baglioni · 2021-11-16T15:23:41+01:00

miriam.baglioni commented

2021-11-16 15:23:41 +01:00

This PR extends the set of enrichment steps by exploiting the relation of type IsParentOf/IsChildOf between organization to create new affiliation relationships between results and organizations.

It works as follows:
Phase 1. Preparation step.
- It stores the couple organization, set of all its parents.
- It stores the couples result, set of known organization it is affiliated to
- It stores the set of leaves: organizations with no children
- It stoires the already existing set of relations between results and organizations

Phase 2. Execution Steps. The process is iterative. Each time it considers a given set of organizations from which to start to create new relationships. The first time it is executed, the starting set is the one of the leaves. It takes all the results associated to the leaves and tries to create affiliation relation with the leaves direct parents. If no relation already exists in the graph connecting the result and the organization a new relation is made. The new relations are used to extend the set of associations between results and organization. In this way a result connected in step i at the organization o can be propagated also to all the parents of o in subsequent steps.
All the direct parent of the set of leaves are selected and considered as the new set of leaves from which to start the propagation again. The process will stop once there is no more organization through which propagate or we have reached a maximum number of iteration (set to 5)

This PR extends the set of enrichment steps by exploiting the relation of type IsParentOf/IsChildOf between organization to create new affiliation relationships between results and organizations. It works as follows: Phase 1. Preparation step. - It stores the couple organization, set of all its parents. - It stores the couples result, set of known organization it is affiliated to - It stores the set of leaves: organizations with no children - It stoires the already existing set of relations between results and organizations Phase 2. Execution Steps. The process is iterative. Each time it considers a given set of organizations from which to start to create new relationships. The first time it is executed, the starting set is the one of the leaves. It takes all the results associated to the leaves and tries to create affiliation relation with the leaves direct parents. If no relation already exists in the graph connecting the result and the organization a new relation is made. The new relations are used to extend the set of associations between results and organization. In this way a result connected in step i at the organization o can be propagated also to all the parents of o in subsequent steps. All the direct parent of the set of leaves are selected and considered as the new set of leaves from which to start the propagation again. The process will stop once there is no more organization through which propagate or we have reached a maximum number of iteration (set to 5)

miriam.baglioni added the

enhancement

label 2021-11-16 15:23:41 +01:00

claudio.atzori was assigned by miriam.baglioni

2021-11-16 15:23:41 +01:00

miriam.baglioni added 8 commits 2021-11-16 15:23:41 +01:00

652114c641 [affiliationPropagation] first try. preparetion

d0ef7d91c5 adding test resource

09f36cffb8 [Enrichment: Propagation through parent-child relationships] First implementation, testing, and wf for propagation of result to organization through semantic relation

2aca6bfa0a mergin with branch beta

b9d124bb7c [Enrichment: Propagation through parent-child relationships] Added counters, and changed constraint to verify if filtering out the relation (from classname = harvested to classid != propagation)

4c70201412 mergin with branch beta