Building a Recommendation Engine: The XELOPES Library - Realtime Data Mining - page 282

// Create initial function values:
createInitialValues();
showValues();
// Do the online learning:
onlineLearning(agent);
showValues();

}

The most complex part is the implementation of RecommEngineEnvironment
which extends RecommEnvironment . Thus, we restrict our attention to the main
elements of this class. Especially, we do not show the implementations of all
methods of RecommEnvironment but just state the internal variables.

/**
* Environment of recommendation engine example. < p >
*
* There are six states S ¼ (1,2,3,4,5,6) and in all
* of these states the same actions may be recommended: A ¼ S.
*/
public class RecommEngineEnvironment extends

RecommEnvironment

{

// ---------------------------
// Variables definitions
// ---------------------------
/** Set of all states. */
private StateSet states ¼ null;

/** Array of all action sets. */
private ActionSet[] actionsets ¼ null;

// --------- Meta Data States/Actions -------- //
/** Item ID attribute of states. */
private CategoricalAttribute itemIDAtt ¼ new Categorica-
lAttribute("itemIDAtt");

/** Meta data of states, i.e. item indexes. */
private MiningDataSpecification metaDataState ¼ null;

/** Item ID attribute of actions. */
protected CategoricalAttribute recoIDAtt ¼ new Categori-
calAttribute("recoIDAtt");

/** Meta data of actions, i.e. rule indexes. */
private MiningDataSpecification metaDataAction ¼ null;

/** Hashtable of step number. */