Semantics

An exciting new area of MarkLogic 7 is the internal provision of an Enterprise class triple store. MLJS has had widgets to expose this functionality since the Early Access programme in May 2013. In this lesson you'll build a query page with clickable results. You'll also make a click on one widget redraw a graph explorer widget to show the selected Subject's facts.

• The semantic context and triple config objects
• Loading semantic facts with MLJS
• A basic semantic search page
• Exploring subjects
• Configuring your own ontology
• Extension exercise

The semantic context and triple config objects

The search context object helps link widgets together around the same content search. Semantic context does the same for semantic (sparql) queries. Likewise, the tripleconfig object helps define the ontology used and how subjects relate to one another.

I've also started creating a sparql query builder object. This is in the very early stages though so I'm not covering it in this class - we'll instead use the visual sparql query builder which is what most people will need to use anyway.

Triple config at the moment has chainable methods to define classes of semantic Subjects, their RDF types, facts they may have, and potential relationships between them and other objects. This class is used mainly by visual widgets so they can provide human readable names to classes and relationships, and extract the 'name' field (if such a thing exists) for a Subject.

This means you don't always have to show a complex IRI string for the Subject name. Basically it makes working with ontologies much more user friendly and intuitive.

Loading semantic facts with MLJS

First though you'll need some facts. Copy the content of samples/010-load-facts.js in to /src/js/page-loadfacts.js, and create a HTML page called loadfacts.html that references it.

Now have a look inside your loadfacts JavaScript code.

You'll notice this tries to load a named graph using the W3C graph store protocol support of MLJS Core. If the named graph does not exist (i.e. the method errors) then the graph is created with the specified facts. Actually three named graphs are created.

You'll notice also that the saveGraph methods takes arrays of MLJS' own helper JSON format. This is to make getting used to using ontologies simple for everyone. In future I'll also provide methods for importing facts from RDF and other formats.

Save and deploy these files, and visit the loadfacts.html page to load the facts in to your content database.

Now visit /v1/graphs/things in your browser to see the default OOTB mechanism provided by MarkLogic to browse through subjects. Useful for a small amount of facts, but not that user friendly. Let's improve on that.

A basic semantic search page

Create yourself a new page called semantics.html with a page-semantics.js file. Ensure they are linked as usual.

Now use the code below to provide an area for querying and viewing the results of the triple store:-

Copy the following JavaScript in to your JS file:-

  window.onload = function() {
    var db = new mljs();
    db.logger.setLogLevel("debug");

    var error = new com.marklogic.widgets.error("errors");

    try {

      var semctx = db.createSemanticContext();

      var tc = semctx.getConfiguration(); // get triple config object
      tc.addTest(); // now done via semantic config builder functions

      var tripResults = new com.marklogic.widgets.sparqlresults("results");
      semctx.register(tripResults);

      var trip = new com.marklogic.widgets.sparqlbar("query");
      semctx.register(trip);

    } catch (err) {
      error.show(err.message);
    }
  };

You'll need to include the widget-triples and widget-explore JS files in your HTML.

Now save and deploy this page and code. Note that you can select some objects that exist in the drop downs. This is because MLJS by default ships with support for some internal and test ontologies.

Execute a search to see a list of subjects and human readable class and instance names be shown. A good starting place is to just return all Person subjects.

Exploring subjects

Add the following code to your JS page file after the last semctx.register line:-

      var explorer = new com.marklogic.widgets.graphexplorer("explorer");
      tripResults.iriHandler(function(iri) {explorer.drawSubject(iri,null,1,1)});

This code instructs the graph explorer to show the details for the result clicked in the sparql results list.

Save, deploy and retest the page.

You'll notice the Subject in the results list is navigable now, and that the graph explorer shows the facts and related subjects.

Configuring your own ontology

The triple config objects comes with many methods with which to configure your own ontology. Examples of how to use these can be found in the mljs.js file.

To view relevant sections, go to the Core API documentation. Select the tripleconfig class. Now click on one of addFoaf, addFoafPlaces, addMarkLogic, addMovies, addPlaces functions. Under each is a link to the mljs.js file that shows the relevant code.

Read the code sections to understand what each method call does. Below is a summary table:-

Extension exercise

Use a test loading page and mljs.saveGraph to save a set of your own triples that don't use any of the existing RDF Types.

Now reconfigure the test page's tripleconfig object to support your ontology.

Course Home Page | Next - 8. Lunch