search

opendatatrentino / OpenDataRise

Tool to cleanse and semantify datasets from CKAN repositories. Based on OpenRefine.
Other
23 stars 3 forks source link

Implement proper Ref to tree schema matching #103

Open DavidLeoni opened 9 years ago

DavidLeoni commented 9 years ago

Current schema matching in ODR supports mapping from column names to etype attributes at most 2 levels deep, and let's admit it code sucks.

To improve things, we should implement a proper Ref to tree schema matching, where Ref could be DcatRef or ColumnRef and targets could be proper tree paths of etype attribute definitions. at most 2 level deep.

DavidLeoni commented 9 years ago

More comment on 0.3.0 status:

  1. OpenDataSchemaMatcher sends to ODR at most two levels deep attributes (only for STRUCTURE and not ENTITY)
  2. SemanticMatchingServiceController in ODR sends to browser schema correspondences with parent correspondences which are messed up :-/ - correspondances are NOT persisted on ODR server. This causes several problems when refreshing browser page.
  3. browser just ignores them and picks info only from Schema mappings and Schema.getAllAttrCorrs() .

Dirty as hell yet somewhat works.