Virtuoso HTML+Variants Cartridge Configuration Overview The HTML and Variants extractor cartridge handles various XML document formats including HTML, XHTML and pure RDF (XML, Turtle, N3). RDF sources are ingested directly. Sources that look like HTML are passed through HTML Tidy before further extraction can take place. Many dialects for embedding microdata (collectively known as "data-islands") are supported. Certain well-known domain-specific standards (such as Twitter Cards) are also supported. Core HTML At its core, the HTML+Variants cartridge identifies the following aspects of a webpage: Title and other well-known data in <meta> tags such as description, author, keywords, encoding, etc images referenced links to alternate forms of the current document links to other referenced documents x509 Certificate fingerprints the page content any RSS or Atom feeds it finds In a second extraction, the following are also supported: Twitter Cards OpenGraph markup (og: prefix) any link@rel becomes a predicate any meta@name becomes a predicate Because the mapping between an HTML document and a logical primary entity is not always clear, there is an option whereby any mention of latitude and longitude in a meta tag is extracted Data Islands There are many ways in which an HTML document can contain embedded data, termed data-islands. Because of its significance, HTML5 Microdata is regarded as an island, although it is processed along with other formats in the GRDDL loop - a second island encompassing: AB Meta BSBM Dublin Core eRDF geoURL Google Base hAudio hCalendar hCard hListing hNews hProduct hRecipe hResume hReview hReview-aggregate HR-XML Microsoft XML Spreadsheet 2003 Ning Metadata relLicense Slidy Word 2003 XML Document XBRL XDDL XFN Profile XFN Profile2 xFolk Recent W3C standards have proposed the use of the HTML <script type="text/turtle">...</script> tag for storing Turtle RDF triples. Relatedly, Google promote the use of the same HTML <script type="application/ld+json">...</script> tag for storing data in JSON-LD format. Reification Data in each of the above islands can be reified (default: enabled apart from the plethora of GRDDL microformats), by which for every statement extracted, an rdf:Statement entity is created summarizing the statement's subect, predicate and object/value along with a pointer to the data-island whence it came. Enabled by default, this reification can be disabled selectively for HTML core + meta extraction (together) RDFa data Turtle-in-script JSON-LD-in-script all the supported GRDDL formats HTML5-Microdata Passthrough Mode Previously, the HTML+Variants cartridge used to be regarded as a fall-back, to be invoked in cases where no other extractor cartridge could handle the document. This has changed; by default, the HTML cartridge is now the first to be invoked; this requires its exit-status be set to permit control to flow into other cartridges. If the HTML cartridge detects another enabled cartridge that might contribute to the data, it assigns its output to the container entity representing the HTML document; in cases where no other cartridge is likely, it makes statements using the primary entity URI as a subject instead. If you wish to move the HTML cartridge back to the bottom of the list, you should also change the passthrough back to `no'. Options The HTML+Variants cartridge supports the following options: passthrough_mode=yes - whether to expect other cartridges to contribute fallback-mode=no - whether to run only if no other cartridge has contributed get-feeds=no - whether to include data from RSS and Atom feeds preview-length=512 (deprecated) - how long a bibo:abstract to generate add-html-meta=yes - add data extracted from <meta> tags rdfa=yes - include data from RDFa reify_html=1 - whether the core data should be reified reify_html5md=1 - whether the HTML5-Microdata should be reified reify_rdfa=1 - whether RDFa data should be reified reify_turtle=1 - whether Turtle data embedded in <script> elements should be reified reify_jsonld=1 - whether JSON-LD data embedded in <script> elements should be reified reify_all_grddl=0 - whether all the GRDDL sources should be reified loose=yes - look for latitude and longitude information in meta tags