TwIdw—A Novel Method for Feature Extraction from Unstructured Texts

This research proposes a novel technique for fake news classification using natural language processing (NLP) methods. The proposed technique, TwIdw (Term weight–inverse document weight), is used feature extraction and based on TfIdf, with the term frequencies replaced by depth of words in documents. effectiveness compared to another method—basic TfIdf. Classification models were created random forest feedforward neural networks, within those, three different datasets used. network method KaiDMML dataset showed an increase accuracy up 3.9%. was not as successful only (1%). network, other hand, all datasets. Precision recall measures also confirmed good results, particularly method. has potential be various NLP applications, including problems.