diff --git a/spec/index.html b/spec/index.html
index 6fc5d20..591378d 100644
--- a/spec/index.html
+++ b/spec/index.html
@@ -1164,23 +1164,63 @@ <h5>DELETE WHERE</h5>
<http://example/william> foaf:mbox <mailto:bill@example> .</pre>
</section>
</section>
+
<section id="load">
<h4>LOAD</h4>
- <p>The <code>LOAD</code> operation reads an RDF document from a IRI and inserts its triples into the specified graph in the Graph Store.
- The specified destination graph SHOULD be created if required; again, implementations providing an update service over a fixed set of
- graphs MUST return with failure for a request that would create a disallowed graph. If the destination graph already exists, then no data in that graph will be
- removed.</p>
- <pre class="defn nohighlight"><strong>LOAD</strong> ( <strong>SILENT</strong> )? <em><a data-cite=
- "SPARQL12-QUERY#rIRIREF">IRIref_from</a></em> ( <strong>INTO GRAPH</strong> <em><a data-cite="SPARQL12-QUERY#rIRIREF">IRIref_to</a></em> )?</pre>
- <p><em>IRIref_from</em> specifies the IRI of a document such that a store will be able to identify, locate and read the document. The most common form will be URLs with the <em>http</em>
- IRI schemes. Once the document has been read, the resulting triples will be inserted into the destination graph named by the IRI referred to by <em>IRIref_to</em>.</p>
- <p>If no destination graph IRI (<em>IRIref_to</em>) is provided to load the triples into, then the data will be loaded into the default graph.</p>
- <p>In case no RDF data can be retrieved (as opposed to the empty graph being retrieved) from the IRI denoted by <em>IRIref_from</em>, or in case the retrieval method returns an error (such
- as, for instance an HTTP error code), the SPARQL 1.2 Update service SHOULD return failure and the status of the Graph Store
- SHOULD remain in the same status as prior to the request; in case the keyword SILENT is present, however, the operation will
- still return success and the status of the Graph Store is not specified by the present document: implementations may create the destination graph or not and partially load data, in case of
- a transmission error where partial data has been received (which itself may be legal RDF).</p>
+ <p>
+ The `LOAD` operation reads an RDF document from an IRI and inserts triples
+ from the document into the <a href="#defn_graphStore">Graph Store</a>.
+ </p>
+
+ <pre class="defn nohighlight"><strong>LOAD</strong> ( <strong>SILENT</strong> )? <em><a
+ data-cite="SPARQL12-QUERY#rIRIREF">IRIref_from</a></em> ( <strong>INTO GRAPH</strong> <em><a
+ data-cite="SPARQL12-QUERY#rIRIREF">IRIref_to</a></em> )?</pre>
+
+ <p><em>IRIref_from</em> specifies the IRI of a document, such that a store will be able to
+ identify, locate, and read the document. The most common IRIs will be URLs in
+ the `https` and `http` IRI schemes.
+ </p>
+
+ <p>If a destination graph (<i>IRIref_to</i>) is specified by `INTO GRAPH`, then the
+ specified destination graph SHOULD be created if required; again, implementations
+ providing an update service over a fixed set of graphs MUST return with failure for a
+ request that would create a disallowed graph. If the destination graph already exists,
+ then no data in that graph will be removed.
+ The RDF document is retrieved and parsed to produce an
+ <a data-cite="RDF12-CONCEPTS#dfn-rdf-graph">RDF Graph</a>.
+ This graph is
+ <a data-cite="RDF12-SEMANTICS#dfn-merging">merged</a> into the destination
+ graph in the graph store.
+ </p>
+ <p>
+ If no destination graph IRI is specified by `INTO`, then the RDF document is retrieved
+ and parsed into an <a data-cite="RDF12-CONCEPTS#dfn-rdf-dataset">RDF Dataset</a>.
+ This dataset is <a data-cite="RDF12-SEMANTICS#dfn-rdf-dataset-merge">merged</a> into the
+ <a href="#defn_graphStore">graph store</a> viewed as a mutable
+ <a data-cite="SPARQL12-QUERY#sparqlDataset">RDF Dataset</a>.
+ </p>
+
+ <p>
+ If no RDF data can be retrieved (as opposed to an empty graph being retrieved) from the
+ IRI denoted by <em>IRIref_from</em>, or in case the retrieval method returns an error
+ (such as, for instance an HTTP error code), the SPARQL 1.2 Update service SHOULD return
+ failure and the status of the Graph Store SHOULD remain in the same status as prior to
+ the request. However, if the the keyword `SILENT` is present, operation will return
+ success, rather than failure, and the status of the Graph Store is is not specified.
+ </p>
+
+ <div class="note" id="context-triples">
+ <p>
+ Some RDF syntaxes, such as [[[JSON-LD11]]], are often used to serialize
+ a single graph with no graph name.
+ [[[RDF12-N-QUADS]]] may be used as a
+ database dump format for a triple store with the data only in the default graph.
+ When the `LOAD` operation has an `INTO GRAPH` clause, it is an error if
+ named graphs are encountered when parsing the RDF document.
+ </p>
+ </div>
</section>
+
<section id="clear">
<h4>CLEAR</h4>
<p>The <code>CLEAR</code> operation removes all the triples in the specified graph(s) in the Graph Store.</p>
@@ -1564,12 +1604,34 @@ <h4>Delete Insert Operation</h4>
<h4>Load Operation</h4>
<div class="defn">
<p><strong><span class="doc-ref" id="defn_loadOperation">Definition: Load Operation</span></strong></p>
- <p>A <a href="#load">Load Operation</a> OpLoad is an Update Operation in which new triples (from a remote graph) are added in the Graph Store, either in the default slot or in a named
- slot, if specified.</p>
- <p>OpLoad ( GS, <em>documentIRI</em>) = Dataset-UNION ( GS, { graph ( <em>documentIRI</em> ) } )</p>
- <p>OpLoad ( GS, <em>documentIRI</em>, <em>iri</em> ) = Dataset-UNION(GS, { {}, ( <em>iri</em>, graph ( <em>documentIRI</em> ) ) } )</p>
- <p>where graph(<em>documentIRI</em>) is a function returning the RDF graph serialized by the RDF document retrieved from IRI <em>documentIRI</em>, where blank nodes present in the
- retrieved graph are supposed to be "standardized apart"; i.e., blank nodes from a loaded graph need to be disjoint with the blank nodes already present in the Graph Store GS.</p>
+ <p>
+ A <a href="#load">Load Operation</a> `OpLoad` is an Update Operation in which new
+ triples (from a remote graph or dataset) are added to the Graph Store, either in the
+ default slot or in a named slot, if specified.
+ </p>
+
+ <p>OpLoad ( GS, <em>documentIRI</em>) = dataset-merge ( GS, dataset ( <em>documentIRI</em> ) } )</p>
+
+ <p>OpLoad ( GS, <em>documentIRI</em>, <em>iri</em> ) = dataset-merge(GS, { {}, ( <em>iri</em>, graph ( <em>documentIRI</em> ) ) } )</p>
+
+ <p>where</p>
+ <ul>
+ <li>The operation <code>dataset ( <em>documentIRI</em> )</code> is a function
+ returning the RDF dataset serialized by the RDF document</li>
+ <li>The operation <code>graph ( <em>documentIRI</em> )</code> is a function
+ returning the RDF graph serialized by the RDF document</li>
+ <li>
+ The operation `dataset-merge` combines the Graph Store GS with the given
+ dataset using the
+ <a data-cite="RDF12-SEMANTICS#dfn-rdf-dataset-merge">dataset merge operation</a>
+ </li>
+ </ul>
+ <p>
+ Blank nodes present in the retrieved graph or dataset are
+ "standardized apart"; i.e., blank nodes from a loaded graph or
+ dataset are disjoint from the blank nodes already present in the
+ Graph Store GS.
+ </p>
</div>
</section>
<section id="def_clearOperation">
@@ -1842,6 +1904,7 @@ <h2>Changes between SPARQL 1.1 Update and SPARQL 1.2 Update</h2>
<li>Add reifying triples examples in <a href="#insertData" class="sectionRef"></a>, <a href="#deleteData" class="sectionRef"></a>, and <a href="#delete" class="sectionRef"></a></li>
<li>Use PREFIX instead of @prefix in all Turtle examples</li>
+ <li>Revise `LOAD` description and definition for handling RDF documents which contain a dataset</li>
</ul>
</section>