2b:null 2c:["$","title",null,{"children":"About RDF annotation | James Anderson | Observable"}] 31:Tc5d,md` There is interest to emulate the expressive capabilities of property graphs in RDF forms. Attention focuses on an approach called "RDF *", which extends RDF forms with triple forms as subject and object terms in order to express RDF statement annotations that are equivalent to link attributes in property graphs. Somehow, the [discussion](https://blog.liu.se/olafhartig/2019/01/10/position-statement-rdf-star-and-sparql-star/) neglects the aspects of [N3](https://www.w3.org/TeamSubmission/n3/) to which it closely relates. Hayes did [mention](https://lists.w3.org/Archives/Public/public-rdf-star/2019Aug/0001.html) that in passing, but the thread was inexplicably neglected. He also [called attention](https://lists.w3.org/Archives/Public/public-rdf-star/2019Sep/0052.html) to issues from treating "entailed assertion (PG) and "separate assertion" (SA) semantics as representations associated with syntax (variants) rather than as independent interpretations and reiterated the insight, that PG interpretations are possible based on an abstract syntax which permits SA interpretations, while the opposite is not. That also, inexplicably, carried little weight, as the mode distinctions remain a topic of concern for implementors. The discussion also neglects DuCharme's [observation](http://www.snee.com/bobdc.blog/2018/04/reification-is-a-red-herring.html), that situations which require attribution are likely best accommodated by reifying the intended class, not just the single statement. Hartig [contradicts]() this, but without motivation. It is argued, that the imperative for a compact reification syntax follows from its succinctness ([hartig.2019,p2](https://www.diva-portal.org/smash/get/diva2:1141963/FULLTEXT01.pdf)) and from the limitations which an abstract syntax involving named graphs would place on "an application of named graphs for other use cases." Booth [called out](https://lists.w3.org/Archives/Public/public-rdf-star/2020Sep/0046.html) the equivalence to singleton names graphs, but that was [rejected](https://lists.w3.org/Archives/Public/public-rdf-star/2020Sep/0047.html) - without substantive justification, on the basis that > annotations on the level of individual triples are something different then annotations on the level of a set of triples (at least for me). For the latter we have things such as N3, and RDF* is for the former. This despite that [Dorte and Van Woensel](https://biblio.ugent.be/publication/8632551/file/8653457.pdf) have described how secondary assertions with regard to the interpretation of referenced graphs are sufficient for N3 to subsume RDF* without respect to referenced triple set cardinality. The discussion also proposes that the referenced triples cannot be comprehended as named graphs, because they ["have no name"](https://lists.w3.org/Archives/Public/public-rdf-star/2019Sep/0068.html). This may be true, but they still have identity, which means, it matters that in the following \`\`\` <:s :p :o> :p2 :o2; :p3 :o3 - <:s :p :o> :p4 :o4 . \`\`\` the \`<:s :p :o>\` forms designates one triple on the first line, but may or may not also designate the same triple on the second. `32:T432,md`## Alternative Implementations There are several provisional implementations: - rdf4j : https://rdf4j.org/documentation/programming/rdfstar/ Extends the term data model with a triple term type. - Stardog : They suggest that "edge properties" support finer granularity than modeling the assertions with named graphs (https://www.stardog.com/blog/property-graphs-meet-stardog/) - Virtuoso : ? - Neptune : ? - GraphDB : The "meaning is aligned to the semantics of standard reification." The describe, on one hand, that the encoding serves just as a more effecient serialization, but on the other, they extend the schematic storage model with an additional type follow the "separate assertion" (SA) approach to schematic storage of the embedded triples. (http://graphdb.ontotext.com/documentation/9.2/free/devhub/rdf-sparql-star.html) - AllegroGraph [alludes](https://franz.com/agraph/support/documentation/current/triple-attributes.html) to their facility for attaching attributes to triples as the implementation path. This implies just triples as the target. `33:T6b3,md`## Dydra Implementation In order to investigate how these capabilities might interact with Dydra's implementation, we investigate how to express the required relationships with as little change as possible to the ways the service implements the RDF abstract model. One consequence is that we do not extend the RDF model to permit triples as terms, but instead explore how to provide the capabilities with N3 formulae. Our sentiment is, for the effort required to augment the media codecs, and to extend the representation and or interpretation semantics, it would be worthwhile to support annotated graphs. We follow Dorte et.al to represent statement annotations as graphs, which we treat as N3 formulae, but as work on standardizing N3 is ongoing, there is no standard approach to using rules, to govern their interpretation. Instead, we store them as content-addressed graphs, for which the respective hash value identifies the graph and interpret the graphs according to how the data set combines them in the FROM clause. This permits both "separate assertion" and "property graph" semantics. - \`\` treats all graphs as constituents of the default graph. This is provides PG semantics. - \`\` leaves each named graphs distinct, but addressible, but eliminates the default graph. - no designation leaves the annotations in distinct graphs and any default graph distinct. This provides SA semantics. Additional dataset designators could be added, for example, to distinguish annotated graphs from other named graphs. A formula identity is computed by canonicalizing the graph, computing its SHA256 hash and encoding that as a URN of the form \`\`. `34:[] 2f:{"id":"7f9629bf197ebf02","slug":"about-rdf-annotation","title":"About RDF annotation","thumbnail":"0e1bb1c513275e69706619ef60febfd30e4087d801ac6e45e0fa8da694322287","access_level":"public","create_time":"$D2020-06-12T00:59:14.178Z","update_time":"$D2024-10-02T14:38:10.761Z","view_time":null,"first_public_time":"$D2020-10-18T09:42:40.004Z","fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"b452fd69977f5701","login":"lomoramic","name":"James Anderson","avatar_url":"https://avatars-next.observableusercontent.com/dc520bbc733259f2913522cf1d50607777b98a3083f290a6f182534561896497","tier":"starter_2024"},"version":663,"publish_version":null,"role":null,"authored":false,"authors":[{"type":"individual","id":"987c90ad70c449c1","login":"lomoramic","name":"James Anderson","avatar_url":"https://avatars-next.observableusercontent.com/dc520bbc733259f2913522cf1d50607777b98a3083f290a6f182534561896497","tier":null}],"members":[{"type":"individual","id":"987c90ad70c449c1","login":"lomoramic","name":"James Anderson","avatar_url":"https://avatars-next.observableusercontent.com/dc520bbc733259f2913522cf1d50607777b98a3083f290a6f182534561896497","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"subscribed":null,"forks":2,"collection":null,"type":"notebook","body":{"title":"About RDF annotation","cells":[{"id":611,"value":"md`# About RDF annotation\n## Formulae and RDF*\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":0,"value":"$31","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":595,"value":"$32","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":605,"value":"md`## Standardization\n\nThere is an effort within a W3C community group to [define](https://w3c.github.io/rdf-star/) what RDF-star is to mean.\nTheir work has led to a [draft](https://w3c.github.io/rdf-star/cg-spec/2021-02-18.html) report.\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":599,"value":"$33","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":22,"value":"md` A simple annotation of a formula abstracted over the variable \\`?o\\` would be \n\n\\`\\`\\`\nselect *\nwhere {\n {_:f1 0 ;\n ?o }\n 1 .\n}\n\\`\\`\\`\n\nThat parses as an outer BGP with the formula present as an inner BGP in the subject position of the initial triple.\nThe abstract SPARQL algebra expression would be\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":385,"value":"{ var algebra = await getQueryAlgebra(`select *\nwhere {\n {_:f1 0 ;\n ?o }\n 1 .\n}`)\nreturn html`

${algebra}

`;\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":296,"value":"md`\nThe example includes an abstract formula as an indication of the use in a SPARQL query expresssion, while\nRDF documents, on the other hand permit literal formulae only.\nThe formula syntax permits multiple constituents.\nWhere the constuent is limited to a single triple, the syntax is logically equivalent to SPARQL-star and\nmay be represented with the '\\`\\`<<\\`\\` \\`\\`>>\\`\\`' delimiters proposed by that external syntax.\n\n[nb, the example result elides urls. they are apparent in the source only.]\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":332,"value":"md`## Examples\n\nThe examples below indicates simple query expressions for each of the variations among subject or object formulae and constant or abstract forms.\n\nThis initial set concerns just the relationships captured in the abstract algebra.\nA later set will depict the consequences for the query plan.\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":334,"value":"generateAlgebraExample(\"constant subject formula\",\n `select *\nwhere {\n {_:f1 0 }\n 3 .\n}`)\n","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":339,"value":"generateAlgebraExample(\"abstract subject formula\",\n `select *\nwhere {\n {_:f1 0 ;\n ?o }\n 1 .\n}`)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":341,"value":"generateAlgebraExample(\"constant object formula\",\n `select *\nwhere {\n \n {_:f1 0 }.\n}`)\n","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":343,"value":"generateAlgebraExample(\"abstract object formula\",\n `select *\nwhere {\n \n {_:f1 ?o }.\n}`)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":345,"value":"generateAlgebraExample(\"heterogeneous subject and object formulae\",\n `select *\nwhere {\n {_:f1 ?o }\n \n {_:f2 0 }.\n}`)\n","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":347,"value":"generateAlgebraExample(\"a nested constant subject formula\",\n `select *\nwhere {\n {_:f1 \n {_:f3 2 } }\n \n {_:f2 0 }.\n}`)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":349,"value":"generateAlgebraExample(\"a nested abstract subject formula\",\n `select *\nwhere {\n {_:f1 \n {_:f3 ?o } }\n \n {_:f2 0 }.\n}`)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":515,"value":"generatePlanExample(\"execution plan for a nested abstract subject formula\",\n `\nselect ?what\nfrom \nwhere\n{ \n { a ?what }\n}\n`)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":634,"value":"generateAlgebraExample(\n \"simple annotation\",\n `\nselect *\nwhere {\n \n{ ?o }\n}\n`\n)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":637,"value":"pathAnnotation = generateAlgebraExample(\n \"path annotation\",\n `\nprefix star: \nselect *\nwhere {\n star:p/star:q \n{ ?o }\n}\n`,\n \"nb. The relevant BNF is not in accord with the report's syntax, as it effectively modifies the pattern for \\`TriplesSameSubject\\` to be \\`(TriplesSameSubject Annotation?\\` rather than modifying the \\`ObjectList\\`pattern to reflect \\`Object Annotation?\\`, as the former suffices for this investigation and is easier to model during the reduction process. This supports annotating the star graph respective the given subject, rather than the single path from that subject. \"\n)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":645,"value":"kleeneAnnotation = generateAlgebraExample(\n \"kleene star annotation\",\n `\nprefix star: \nselect *\nwhere {\n star:p* \n{ ?o }\n}\n`\n)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":256,"value":"md`\n## Appendix\n`\n","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":204,"value":"import { resourceAPI } from \"@datagenous/resource-protocol-library\"","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":217,"value":"sparqlPost = resourceAPI(\"https://dydra.com/test/test\", \"POST\")","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":394,"value":"function getQueryAlgebra(expression) {\n return ( sparqlPost(expression, {\"Accept\": \"application/sparql-query-algebra\", \"Content-Type\": \"application/sparql-query\"}) );\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":513,"value":"function getQueryExecutionPlan(expression) {\n return ( sparqlPost(expression, {\"Accept\": \"application/sparql-query-execution\", \"Content-Type\": \"application/sparql-query\"}) );\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":272,"value":"function generateAlgebraExample(description, query, comment = null) {\n var p = document.createElement('div');\n p.style.borderTop = \"solid 1px black\";\n var span1 = document.createElement('span');\n span1.innerText = description;\n span1.style.fontWeight = \"bold\";\n p.appendChild(span1);\n var pre1 = document.createElement('pre');\n pre1.innerText = query;\n p.appendChild(pre1);\n var span2 = document.createElement('span');\n span2.innerText = \"would yield the algebra\";\n p.appendChild(span2);\n var pre2 = document.createElement('pre');\n getQueryAlgebra(query).then(function(text) {\n pre2.innerText = text;\n });\n pre2.innerText = \"\";\n p.appendChild(pre2);\n if (comment) {\n var div = document.createElement('div');\n div.innerText = comment;\n p.appendChild(div);\n }\n return p;\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":517,"value":"function generatePlanExample(description, query) {\n var p = document.createElement('div');\n p.style.borderTop = \"solid 1px black\";\n var span1 = document.createElement('span');\n span1.innerText = description;\n span1.style.fontWeight= \"bold\";\n p.appendChild(span1);\n var pre1 = document.createElement('pre');\n pre1.innerText = query;\n p.appendChild(pre1);\n var span2 = document.createElement('span');\n span2.innerText=\"would yield the plan\";\n p.appendChild(span2);\n var pre2 = document.createElement('pre');\n getQueryExecutionPlan(query).then(function(text) { pre2.innerText=text;});\n pre2.innerText = \"\";\n p.appendChild(pre2);\n return (p);\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":505,"value":"md`\n## Notes\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":499,"value":"md`\nThis document directs its requests to the Dydra service at \\`dydra.com\\`. \nIf one wishes to experiment, please note, the annotation support in that host is a development capability and subject to change.\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":553,"value":"md`\n### examples`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":538,"value":"getQueryExecutionPlan(`\nselect ?what\nfrom \nwhere\n{ \n { a ?what }\n}`)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":550,"value":"sparqlPost(\n `select ?what\nfrom \nwhere\n{ \n { a ?what }\n}`,\n {\n Accept: \"application/sparql-query-algebra\",\n \"Content-Type\": \"application/sparql-query\"\n }\n)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":250,"value":"\n sparqlPost(\"select * where {?s ?p ?o4}\", {\"Accept\": \"application/sparql-results+json\", \"Content-Type\": \"application/sparql-query\"});","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":620,"value":"import { catalog } from \"@lomoramic/blog-catalog\"","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"}],"files":"$W34","stdlib":"1","titling":"auto"}} 30:"$2f:body"