The Semi-Supervised LearningThe Semi-Supervised Learning (SSL) paradigm is an extension of supervised learning with the principle of inclusion of unlabeled in- stances, which are used as background knowledge. When using SSL for a classification task, the entire approach is usually denoted as a Semi-Supervised Classification (SSC) ( Chapelle et al., 2010 ), where unlabeled data can be understood as part of the... (SSL) paradigm is an extension of supervised learning with the principle of inclusion of unlabeled in- stances, which are used as background knowledge. When using SSL for a classification task, the entire approach is usually denoted as a Semi-Supervised Classification (SSC) ( Chapelle et al., 2010 ), where unlabeled data can be understood as part of the entire training set. In general, SSC is intended to learn a better classifier by in- cluding unlabeled instances rather than using only a labeled set. This approach can be performed in two different ways: as trans- ductive learning, where labeled data is used to build a classifier which is then used on unlabeled data; and inductive learning, where the evaluation is carried out on a validation set (a part of data that is excluded from the available training data before the semi-supervised training phase) ( Chen & Wang, 2011 ).

 

The complete paragraph above is taken from:

Pavlinek, M., & Podgorelec, V. (2017). Text classification method based on self-training and LDALDA is a generative probabilistic topic model. It represents the documents as a random mixtures of topics over the latent topic space, where each topic is characterized by a distribution over a dictionary of words. LDA and its extensions are ineffective when used with short documents (texts). Issues are coming from: ineffective word relation induction and difficulties with distinguishing ambiguous... More topic models. Expert Systems with Applications, 80, 83–93. http://doi.org/10.1016/j.eswa.2017.03.020