Projects/Nepomuk/GraphConcepts

From KDE TechBase

Nepomuk is based on the Semantic Web, and even though the Semantic Web is largely advertised as a triple store, Nepomuk is NOT a triple store. We in fact store quadruples -> Subject - Predicate - Object - Graph. The Graph part of these quadruples is always automatically managed by Nepomuk. Clients rarely (if ever) need to care about the graph.

However, if you still want to know why they are present. Read on.

Introduction to Graphs

The graph is always called the context of the statement. It is used to store some metadata about each triple. In our case we currently store the following information

  • Graph Type
  • Creation Date
  • Modification Date
  • Graph Maintainer


Example -

<nepomuk:/ctx/c4d93812-7d8c-4c8f-b8a7-e1d4dbc5fed5>
        rdf:type                nrl:InstanceBase
        nao:created             2011-07-30T11:34:19.587Z
        nao:modified            2011-07-30T11:34:19.587Z
        nao:maintainedBy        nepomuk:/res/e2eb2efb-14ee-4038-ac24-698f916289b0


The nao:created and nao:lastModified properties are identical to those used in resources. Except for the fact that graphs are generally never modified. They are only created or destroyed.

Graph Type

Each graph has a type which depends on the type of data it holds. The 3 most commonly used types are -

  • nrl:Ontology - Used to hold ontology data.
  • nrl:InstanceBase - Used for normal data
  • nrl:DiscardableInstanceBase - Used for data which can be reproduced and should not be backed up. This type is generally reserved for data created during indexing of files.

Agents

Each graph contains a property called nao:maintainedBy which specifies which application created that data -

<nepomuk:/ctx/c4d93812-7d8c-4c8f-b8a7-e1d4dbc5fed5>
        rdf:type                nrl:DiscardableInstanceBase
        nao:created             2011-07-30T11:34:19.587Z
        nao:modified            2011-07-30T11:34:19.587Z
        nao:maintainedBy        nepomuk:/res/e2eb2efb-14ee-4038-ac24-698f916289b0

<nepomuk:/res/e2eb2efb-14ee-4038-ac24-698f916289b0>
        rdf:type                nao:Agent
        rdf:type                rdfs:Resource
        nao:identifier          nepomukindexer


Each application is represented as a nao:Agent with the application name specified as the nao:identifier. This name is automatically determined via KComponentData and can be spoofed. Security was not one of our concerns.

Each triple may be maintained by a number of different applications. This information is valuable for the advanced DataManagement functions such as removeDataByApplication.