LINKED OPEN METADATA AND ONTOLOGIES

• My name is Riccardo Grosso (riccardo.maria.grosso@gmail.com)

• I am an hobbyst metadata architect • Actually I work at the datacenter of the CSI-Piemonte• For many years I worked in metadata architecture,

first in other companies, then in CSI-Piemonte• This passion has filled a lot of my free time• The first thanks goes to the CSI if I continue to

cultivate this passion with new ideas in my free time• The first ex equo thanks to Carlo Batini & Matteo

Palmonari of Milan-Bicocca University, Roberto Moriondo of Piedmont Region

• My name is Riccardo Grosso. In last century I started working in data and metadata conceptual modeling. My passion is to discover relationships between data, and processes to which they relate, in similar formalisms to natural language. Subject, verb, object complement. Citizen pays tribute. Recently I’ve realized a metadata catalog for a government to classify informatic objects and assets. Particularly, with regard to the metadata of the databases, I have experimented semantics inferences "for similarity" in the descriptive metadata (names and descriptions of tables and fields) to extract and exploit the knowledge embedded in the metadata, using conceptual schema frameworks also called light ontologies .

2

3

• There is no 'time to talk about the project and the experiments about operational and decisional databases, but I think that this work can be reused, to enhance the knowledge in the linked open data world, especially public administration that we have conceptual schemas . Linked open data are in fact tables, with columns, with constraints from columns

LOM

Insert name and descr.of tables & columns.Infer physical constraints

Tag metadata by likenessof entity-names inside in

names and descr. of tables & columns

Integrated &abstractedE-R schemarepository

LOD«Universe»

(data & metadata)

Linked Open Data(RDF/OWL)

Infer verticalAbstraction level

Entity-ierarchytaxonomy

ReuseRelationships

LOMO…

From LOD (Linked Open Data) to LOMO (Linked Open Metadata & Ontologies)

5

• - We can populate the metadata universe of linked open data in public administration - building the data structures that contains the linked open data available - insert names and descriptions of tables and fields of linked open data, and their constraints formed by primary and foreign keys - Use entities 'conceptual frameworks available for similarity to tag where these entities' are present in the names and descriptions of tables and fields (it can be better with the use of text mining, or computational linguistics ) - Reassemble the hierarchy of the entities' tagged - To infer a level of abstraction of the concepts measured vertical - so we obtain linked open metadata level

• But we are not so happy, and we want to relate the entities' extracted, with the relationships present in the light ontologies, and reuse all the repository of integrated and abstracted conceptual schemes . The abstraction helps to simplify and better understand the tagged knowledge. - We then relate the entities extracted - Now we are happy: we have linked open metadata and ontologies, and vertical relations of abstractions, given by hierarchies, and horizontal correlations from entities- The hierarchy of entities' are the steps of ascent in the abstraction of the conceptual schemas, represent the level of abstraction

6

• PROPOSAL:– So we could think: if we create a collaborative

project to make everything?– Administrations should only engages with IT

resources that manage their services , providing logical and physical schemas of linked open data published ( for insiders : most important is the sql – bill with comments of tables and fields , pardon the jargon )

– Such schemes would be instanced by administrations , and cataloged. Most important is to catalog where is each linked open data db implemented in datacenter (physical instance)

– The catalogs get made public.– The metadata catalogs would be enhanced with the

use of ontologies for semantic inference , and conceptual schemes instantiated on each data base structure , with each administration, always in a public way

– And what about the cost ? I personally put my experience, my methods and my vintage tools of inference

• WHY ?– Best reuse of knowledge– Helps in master data management– Ontological query, 2 levels:

• Ontologies and linked open metadata driving:– Physical data query

8