**3. Support vector machine (SVM) model**

A classification model named Support Vector Machine (SVM) is used in this method. Support Vector Machine is a supervised learning model. In its application, several linear functions of high dimensional space (feature space) are utilized. This linear function aims to find the best hyperplane in maximizing each class gap [8].

*Text Classification on the Instagram Caption Using Support Vector Machine DOI: http://dx.doi.org/10.5772/intechopen.99684*

In short, support vector machine is a linear classifier. However, in some nonlinear problems, this model can also be used with some improvements [9], which are needed because not all data is linearly divided. This results in non-optimal results if linear SVM is still applied.

The radial basis function (RBF) kernel was used to change the SVM modeling process from linear to non-linear [10]. Generally, the RBF kernel is used for all types of data as a linear data separator. The RBF kernel has two parameters, namely Gamma and Cost.

The Cost parameter is used for SVM optimization so that misclassification in the training dataset sample nghwaes not occur. Meanwhile, to measure the influence given by each training dataset sample, the Gamma parameter is used [11]. A low or high value is indicated by the use of this parameter. Low or high values are described as "far" and "near". The formula below is the RBF Kernel equation:

$$K\_{(\mathbf{x},\mathbf{z})} = \exp\left[-\boldsymbol{\gamma} \|\mathbf{x} - \mathbf{z}\|^2\right] \tag{4}$$
