- trainul se facea cu statusul functie de ce actiune facea adminul (clean/spam) sau default cand posta userul un comment (active):

var output = googleSpamPrediction.InsertTrainingData('clean/spam/active',
predictionData.Body,
predictionData.NumberOfDaysSinceUserRegistered,
predictionData.NumberOfCommentBefore,
predictionData.NumberOfDistinctTargetUserCommentedOn,
predictionData.NumberOfCommentsWithinAWeek,
predictionData.NumberOfCommentsWithinADay,
predictionData.AvgTextLengthLastFiveComments,
predictionData.DeviationTextLengthLastFiveComments);


- predict se facea cu un apel de genul asta, rezultand simplu - clean/spam:

var output = googleSpamPrediction.Predict(predictionData.Body,
predictionData.NumberOfDaysSinceUserRegistered,
predictionData.NumberOfCommentBefore,
predictionData.NumberOfDistinctTargetUserCommentedOn,
predictionData.NumberOfCommentsWithinAWeek,
predictionData.NumberOfCommentsWithinADay,
predictionData.AvgTextLengthLastFiveComments,
predictionData.DeviationTextLengthLastFiveComments);


Fisierul rezultat... de training.. banuiesc e asta: sins_trainingData_final.zip