Skip to content

Latest commit

 

History

History
442 lines (414 loc) · 15.1 KB

Serialization.md

File metadata and controls

442 lines (414 loc) · 15.1 KB
Raw

Serialization

The definition above describes an abstract data model for a timestamped graph because it is based on an RDF Dataset, which is an abstract entity. The serialization of such timestamped graphs is an additional issue. A timestamped graph could be serialized in an existing format for RDF Datasets, e.g. Trig. However, the Trig syntax does not provide abbreviations for the special cases that arise for a timestamped graph. Therefore, we must introduce a new concrete syntax for timestamped graphs. There are several syntactic categories.

'n' is a blank node.

The intention is that this blank node will not be shared with any other RDF graphs or RDF Datasets. Therefore, it can be represented implicitly, meaning processors will generate a new blank node for each timestamped graph they receive. As an illustration, the serialization could take the form of a space-separated tuple surrounded by parentheses: (g p t). For example, a serialization of a timestamped graph containing two triples would appear as follows

(
  {
       _:b foaf:name "Alice" .
       _:b foaf:mbox <mailto:[email protected]> .
  }
  <http://streamreasoning.org/timeVocab#startsAtTime>
  "2012-11-02T19:14:23+01:00"^^xsd:dateTime
)

When we say "a timestamped graph contains two triples", that is shorthand for "a timestamped graph whose named graph contains two triples" This syntax could potentially be abbreviated further by making some assumptions, e.g.

  • the predicate is always an absolute IRI
  • the timestamp is by default of datatype xsd:dateTime With these assumptions, most of the delimiting characters become unnecessary, and the timestamped graph could be serialized so:
(
  {
       _:b foaf:name "Alice" .
       _:b foaf:mbox <mailto:[email protected]> .
  }
  http://streamreasoning.org/timeVocab#startsAtTime
  2012-11-02T19:14:23+01:00
)

'n' is an IRI and the named graph is included explicitly.

As an illustration, the serialization could take the form of a space-separated qudruple surrounded by parentheses: (n g p t). For example:

(
 <http://example.org/alice>
  {
       _:b foaf:name "Alice" .
       _:b foaf:mbox <mailto:[email protected]> .
  }
  <http://streamreasoning.org/timeVocab#startsAtTime>
  "2012-11-02T19:14:23+01:00"^^xsd:dateTime
)

This could potentially be abbreviated further by making some assumptions in addition to those from the previous case, e.g.

  • the name of the named graph is always an absolute IRI With these assumptions, most of the delimiting characters become unnecessary, and the timestamped graph could be serialized so:
(
http://example.org/alice
  {
       _:b foaf:name "Alice" .
       _:b foaf:mbox <mailto:[email protected]> .
  }
  http://streamreasoning.org/timeVocab#startsAtTime
  2012-11-02T19:14:23+01:00
)

'n' is an IRI and the named graph is obtained by dereferencing that IRI.

As an illustration, the serialization could take the form of a space-separated tuple surrounded by parentheses: (n p t). For example:

(
 <http://example.org/alice>
 <http://streamreasoning.org/timeVocab#startsAtTime>
 "2012-11-02T19:14:23+01:00"^^xsd:dateTime
)

This could potentially be abbreviated further by making the assumptions as in the previous cases. With these assumptions, most of the delimiting characters become unnecessary, and the timestamped graph could be serialized so:

(
http://example.org/alice
http://streamreasoning.org/timeVocab#startsAtTime
2012-11-02T19:14:23+01:00
)

Although there may be further abbreviations possible for these special cases, it is also desirable to be able to extend this syntax to additional cases. For example, consider situations where

  • the timestamp is not in the xsd:dateTime datatype. In this case the full form for a literal could be used, "2012-11-02"^^xsd:date
  • the timestamp is complex, e.g. an interval. In this case, a nested unlabelled blank node could be used.
  • the same named graph is referenced by more than one timestamping triple, with timestamps corresponding to different events. In this case a larger tuple, e.g. (g p1 t1 p2 t2) could be used.

Further abbreviations may be considered after a broader range of extensions has been examined.

Note that each of the concrete examples above describes an RDF Dataset - there is a uniform abstract data model for timestamped graphs, independent of the form taken in the concrete syntax.

JSON-LD Serialization

This section describes a potential JSON-LD serialization for a timestamped graph. We will exemplify the syntax on the following graphs:

G1: http://example.org/graphs/1:

  • Generated: 2015-01-15
  • Recorded: 2015-01-17
<http://example.org/person/Jean-Paul> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://xmlns.com/foaf/0.1/Person> .
<http://example.org/person/Jean-Paul> <http://xmlns.com/foaf/0.1/knows> "http://example.org/person/Alasdair"  .
<http://example.org/person/Jean-Paul> <http://xmlns.com/foaf/0.1/knows> "http://example.org/person/Tara"  .
<http://example.org/person/Jean-Paul> <http://xmlns.com/foaf/0.1/name> "Jean Paul Calbimonte"  .

<http://http://example.org/person/Alisdair> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://xmlns.com/foaf/0.1/Person> .
<http://http://example.org/person/Alisdair> <http://xmlns.com/foaf/0.1/knows> "http://example.org/person/Tara" .
<http://http://example.org/person/Alisdair> <http://xmlns.com/foaf/0.1/name> "Alasdair Gray" .

G2:http://example.org/graphs/2:

  • Generated: 2015-01-16
  • Recorded: 2015-01-17
<http://example.org/person/Tara> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://xmlns.com/foaf/0.1/Person> .
<http://example.org/person/Tara> <http://xmlns.com/foaf/0.1/knows> "http://example.org/person/Alasdair"  ..
<http://example.org/person/Tara> <http://xmlns.com/foaf/0.1/name> "Tara Athan"  .

Graphs are exchanged in 'independent' files/streams

In this case, each graph is a JSON object. Each file can define its own context (shortcuts).

G1:

{
 "@context": {
  "generatedAt": {
    "@id": "http://www.w3.org/ns/prov#generatedAtTime",
    "@type": "http://www.w3.org/2001/XMLSchema#date"
  },
  "recordedAt": {
    "@id": "http://www.w3.org/ns/prov#recordedAt",
    "@type": "http://www.w3.org/2001/XMLSchema#date"
  },
  "Person": "http://xmlns.com/foaf/0.1/Person",
  "name": "http://xmlns.com/foaf/0.1/name",
  "knows": "http://xmlns.com/foaf/0.1/knows"
 },
  "@id": "http://example.org/graphs/1",
  "generatedAt": "2015-01-15",
  "recordedAt": "2015-01-17",
  "@graph":
  [
    {
      "@id": "http://example.org/person/Jean-Paul",
      "@type": "Person",
      "name": "Jean Paul Calbimonte",
      "knows": ["http://example.org/person/Alasdair","http://example.org/person/Tara"]
    },
    {
      "@id": "http://http://example.org/person/Alisdair",
      "@type": "Person",
      "name": "Alasdair Gray",
      "knows": "http://example.org/person/Tara"
    }
  ]
}

See in JSON-LD playground

G2:

 {
 "@context": {
  "generatedAt": {
    "@id": "http://www.w3.org/ns/prov#generatedAtTime",
    "@type": "http://www.w3.org/2001/XMLSchema#date"
  },
  "recordedAt": {
    "@id": "http://www.w3.org/ns/prov#recordedAt",
    "@type": "http://www.w3.org/2001/XMLSchema#date"
  },
  "Person": "http://xmlns.com/foaf/0.1/Person",
  "name": "http://xmlns.com/foaf/0.1/name",
  "knows": "http://xmlns.com/foaf/0.1/knows"
 },
   "@id": "http://example.org/graphs/2",
 "generatedAt": "2015-01-16",
  "recordedAt": "2015-01-17",
  "@graph":
  [
    {
      "@id": "http://example.org/person/Tara",
      "@type": "Person",
      "name": "Tara Athan",
      "knows": "http://example.org/person/Alisdair"
    }
  ]
}

See in JSON-LD playground

Graphs are exchanged in the same file/stream

In this case, several graphs are aggregated in the same file or stream. In such case we could use a shared context to reuse shortcuts between different graphs, and also define 'internal' contexts so that specific shortcuts are defined for a graph.

Array of @graph keywords with shared context

All graphs are expressed in an array inside the default graph. The context (shortcuts) defined in the default graphcan be reused by all graphs.

{
  "@context": {
    "generatedAt": {
      "@id": "http://www.w3.org/ns/prov#generatedAtTime",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "recordedAt": {
      "@id": "http://www.w3.org/ns/prov#recordedAt",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "Person": "http://xmlns.com/foaf/0.1/Person",
    "name": "http://xmlns.com/foaf/0.1/name",
    "knows": "http://xmlns.com/foaf/0.1/knows"
  },
   "@graph":
  [
    {
    "@id": "http://example.org/graphs/1",
    "generatedAt": "2015-01-15",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Jean-Paul",
        "@type": "Person",
        "name": "Jean Paul Calbimonte",
        "knows": ["http://example.org/person/Alasdair","http://example.org/person/Tara"]
      },
      {
        "@id": "http://http://example.org/person/Alisdair",
        "@type": "Person",
        "name": "Alasdair Gray",
        "knows": "http://example.org/person/Tara"
      }
    ]
    },
    {
    "@id": "http://example.org/graphs/2",
    "generatedAt": "2015-01-16",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Tara",
        "@type": "Person",
        "name": "Tara Athan",
        "knows": "http://example.org/person/Alasdair"
      }
    ]
    }
  ]
 }

See in JSON-LD playground

Mixing shared and 'internal' contexts

Besides the shared context, each graph can define own its own context. Duplicate context terms are overridden using a most-recently-defined-wins mechanism. In the following example, we have defined an "example" prefix ( "example": "http://example.org/person/") just for the second graph.

{
  "@context": {
    "generatedAt": {
      "@id": "http://www.w3.org/ns/prov#generatedAtTime",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "recordedAt": {
      "@id": "http://www.w3.org/ns/prov#recordedAt",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "Person": "http://xmlns.com/foaf/0.1/Person",
    "name": "http://xmlns.com/foaf/0.1/name",
    "knows": "http://xmlns.com/foaf/0.1/knows"
  },
   "@graph":
  [
    {
    "@id": "http://example.org/graphs/1",
    "generatedAt": "2015-01-15",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Jean-Paul",
        "@type": "Person",
        "name": "Jean Paul Calbimonte",
        "knows": ["http://example.org/person/Alasdair","http://example.org/person/Tara"]
      },
      {
        "@id": "http://http://example.org/person/Alisdair",
        "@type": "Person",
        "name": "Alasdair Gray",
        "knows": "http://example.org/person/Tara"
      }
    ]
    },
    {
    "@id": "http://example.org/graphs/2",
    "generatedAt": "2015-01-16",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
         "@context": {
           "example": "http://example.org/person/"
         },
        "@id": "example:Tara",
        "@type": "Person",
        "name": "Tara Athan",
        "knows": {
        	"@id": "example:Alisdair"
    	}
  	
      }
    ]
    }
  ]
 }

See in JSON-LD playground

Array of @graph keywords with NO shared context

All graphs are expressed in an array with no shared context.

[
  {
   "@context": {
    "generatedAt": {
      "@id": "http://www.w3.org/ns/prov#generatedAtTime",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "recordedAt": {
      "@id": "http://www.w3.org/ns/prov#recordedAt",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "Person": "http://xmlns.com/foaf/0.1/Person",
    "name": "http://xmlns.com/foaf/0.1/name",
    "knows": "http://xmlns.com/foaf/0.1/knows"
   },
    "@id": "http://example.org/graphs/1",
    "generatedAt": "2015-01-15",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Jean-Paul",
        "@type": "Person",
        "name": "Jean Paul Calbimonte",
        "knows": ["http://example.org/person/Alasdair","http://example.org/person/Tara"]
      },
      {
        "@id": "http://http://example.org/person/Alisdair",
        "@type": "Person",
        "name": "Alasdair Gray",
        "knows": "http://example.org/person/Tara"
      }
    ]
  },
   {
   "@context": {
    "generatedAt": {
      "@id": "http://www.w3.org/ns/prov#generatedAtTime",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "recordedAt": {
      "@id": "http://www.w3.org/ns/prov#recordedAt",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "Person": "http://xmlns.com/foaf/0.1/Person",
    "name": "http://xmlns.com/foaf/0.1/name",
    "knows": "http://xmlns.com/foaf/0.1/knows"
   },
     "@id": "http://example.org/graphs/2",
   "generatedAt": "2015-01-16",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Tara",
        "@type": "Person",
        "name": "Tara Athan",
        "knows": "http://example.org/person/Alisdair"
      }
    ]
  }
 ]

See in JSON-LD playground

Graph 'n' is a blank node

Regardless of wheter graphs are exchanged in the same file/stream or not, the graph name can be a blank node by ommiting the "@id" keyword in the graph definition. The following example shows this feature in the case of aggregating both graphs in a file/stream:


  "@context": {
    "generatedAt": {
      "@id": "http://www.w3.org/ns/prov#generatedAtTime",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "recordedAt": {
      "@id": "http://www.w3.org/ns/prov#recordedAt",
      "@type": "http://www.w3.org/2001/XMLSchema#date"
    },
    "Person": "http://xmlns.com/foaf/0.1/Person",
    "name": "http://xmlns.com/foaf/0.1/name",
    "knows": "http://xmlns.com/foaf/0.1/knows"
  },
   "@graph":
  [
    {
    "generatedAt": "2015-01-15",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Jean-Paul",
        "@type": "Person",
        "name": "Jean Paul Calbimonte",
        "knows": ["http://example.org/person/Alasdair","http://example.org/person/Tara"]
      },
      {
        "@id": "http://http://example.org/person/Alisdair",
        "@type": "Person",
        "name": "Alasdair Gray",
        "knows": "http://example.org/person/Tara"
      }
    ]
    },
    {
    "generatedAt": "2015-01-16",
    "recordedAt": "2015-01-17",
    "@graph":
    [
      {
        "@id": "http://example.org/person/Tara",
        "@type": "Person",
        "name": "Tara Athan",
        "knows": "http://http://example.org/person/Alisdair"
      }
    ]
    }
  ]
 }

See in JSON-LD playground