Improving Feature Representation Based on a Neural Network for Author Profiling in Social Media Texts

Table 17

Obtained results (accuracy, %) for gender classification using SVM classifier on the PAN author profiling 2016 Dutch training corpus under 10-fold cross-validation with vectors extracted only from the training data.

Feature set	Gender
	SVM-NP	SVM-WP
D2V (1-gram)	74.18	73.67
D2V (1 + 2-grams)	76.01	76.78
D2V (1 + 2 + 3-grams)	73.95	75.97