Benefits of HTML5-based Knowledge Graph Deployment: Michael Jackson’s Top 147 Songs

I stumbled upon this post by Vulture, “All 147 Michael Jackson Songs, Ranked From Worst to Best”.

Naturally, this triggered a SPARQL utility demonstration aimed at extracting all 147 songs, while also progressively updating our URIBurner Knowledge Graph.

Step-by-Step Demonstration

Step 1. Using the OpenLink Structured Data Sniffer (OSDS), I found the RDF embedded in the HTML was using terms from, so I used the SPARQL SAMPLE() aggregate function to produce an Entity Relationship Graph overview.

Screenshot Sequence

  1. SPARQL Query Result Page

  2. Results Page, including Query Definition via OSDS

  3. Results Page with Query text display via OSDS

  4. Entity Description page — as a result of clicking a Hyperlink from the SPARQL Query results page

Step 2: The RDF embedded in the Vulture document’s HTML isn’t scoped to each song, so I am going to use SPARQL to extract that information via the embedded YouTube hyperlinks.

Results Page Screenshots

Step 3: URIBurner provides an HTML page that describes the Vulture article. Note that this includes NLP contributions from DBpedia Spotlight, Babel Net, diffbot, dataTXT, and others.

Step 4: In situations where SQL is the preferred query language, our Virtuoso lets you use SPARQL nested in the FROM clause (that is, SPASQL, or SPARQL-within-SQL).

Here is a Live Demo Link — Use vdb for both username and password when challenged.

Then follow-your-nose!