Combining Lexical and Semantic Features for Short Text Classification

Abstract

In this study, we offer a new method for categorizing brief texts by integrating lexical and semantic characteristics. We offer a refined
metric for selecting lexical characteristics and then use a reservoir of prior knowledge that spans the domains of interest to identify
relevant semantic features. When words and meanings are put together, it creates done by assigning varying weights to words in a map
of themes. The number of features is reduced to the number of subjects in this manner. Our classification system, a Support Vector
Machine (SVM), uses Wikipedia articles as training data. Results from our experiments demonstrate that, in comparison to other
strategies for labelling brief texts, our approach is more successful.