Shlukování textů ze sociálních sítí

Abstract

This work aims to create a diverse dataset for the evaluation of social network text clustering and explores different combinations of text embedding methods, dimensionality reduction techniques, and clustering methods. We choose a wide range of the most appropriate evaluation metrics, build an evaluation pipeline, and test the most interesting models.This work aims to create a diverse dataset for the evaluation of social network text clustering and explores different combinations of text embedding methods, dimensionality reduction techniques, and clustering methods. We choose a wide range of the most appropriate evaluation metrics, build an evaluation pipeline, and test the most interesting models.