Label Classification Result correct?

[erelsgl]

Label Classification Result correct?

[erelsgl]

That's a good question. I do not know why "req" is not returned.
It may be worthwhile to use the "explain" feature and see the actual numeric feature-vectors.

[Unrealize]

intentClassifier = new limdu.classifiers.EnhancedClassifier({
	classifierType: TextClassifier,  
	normalizer: limdu.features.LowerCaseNormalizer,
	featureExtractor: WordExtractor  
});
intentClassifier.trainOnline("switch on the fan", "fanon");
intentClassifier.trainOnline("switch off the fan", "fanoff");
console.dir(intentClassifier.classify("switch on the fan pls",4));

and the result:

positive:
fanoff: (4) ["switch+0.80", "the+0.80", "fan+0.80", "bias+0.80"]
fanon: (4) ["on+0.64", "switch+0.14", "the+0.14", "fan+0.14"]

why fan off has a higher number than fan on?