Impact of Preprocessing Using Substitution on the Performance of Selected NER Models - Results

This paper investigates the impact of preprocessing using substitution in word sequences on selected models of named entity recognition. The study is focused on evaluating the results of the performed experiments. It briefly describes the concept of substitution using pseudo words and the methodology used in performing the experiments. Based on the outputs of the experiments, it discusses in detail the implications of substitution on the selected models and provides possible explanations for the results. In the end, conclusions and recommendations for the use of substitution as a preprocessing technique are made based on the observed results.