Classification of the Lithuanian Text of Email Enquieries of an Insurance Company with a Big Number of Customer Categories
Technological Sciences
Karolis Kiaunė
Vilnius Gediminas Technical University, Lithuania
Simona Ramanauskaitė
Assoc. Prof. Dr., Vilnius Gediminas Technical University
http://orcid.org/0000-0003-3195-4280
Published 2019-12-18
https://doi.org/10.21277/jmd.v49i2.235
PDF

Keywords

NLP
text classification
emails
text processing

How to Cite

Kiaunė, K. and Ramanauskaitė, S. (2019) “Classification of the Lithuanian Text of Email Enquieries of an Insurance Company with a Big Number of Customer Categories”, Jaunųjų mokslininkų darbai, 49(2), pp. 52–59. doi:10.21277/jmd.v49i2.235.

Abstract

Natural language processing and classification have been widely used in English-speaking countries. However, analysis and classification of a Lithuanian text is a complex issue and has not been fully implemented. This is due to complexity and peculiarities of the Lithuanian language, so methods appropriate for other languages, are not always appropriate for the Lithuanian language.
Three selected word processing options and their various combinations were used and it was assessed how different and consistent text classification methods are able to classify insurance company customers‘ enquiries sent by email. This study is unique because a great number of methods were used and classification accuracy of a Lithuanian text in a large number of categories (33) was further assessed.
Natural language processing problems, analogous studies of Lithuanian text classification were analyzed, research methodology was proposed and research findings were discussed in the paper.

PDF

Downloads

Download data is not yet available.