Word-Graph vs. bag-of-words feature extraction for solving author identification problem

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27510%2F19%3A10243815" target="_blank" >RIV/61989100:27510/19:10243815 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Word-Graph vs. bag-of-words feature extraction for solving author identification problem

Popis výsledku v původním jazyce

In this paper we examine multiple methods for solving the problem of text vectorization in context of text classification. We compare two variants of a traditional Bag-of-Words technique to a newly proposed Word-Graph approach based on graph representation of a text document and measuring similarities between graph structures. We further propose modifications to the Word-Graph method potentially improving classification accuracy. Results of experiments performed while solving an author identification problem on a dataset consisting of speeches made during meetings of Slovak National Parliament show that the Word-Graph approach offers similar levels of accuracy as traditional methods. Proposed modifications significantly improve the performance in case of imbalanced number of documents for each class in the training set. (C) 2019 VSB-Technical University of Ostrava. All rights reserved.

Název v anglickém jazyce

Word-Graph vs. bag-of-words feature extraction for solving author identification problem

Popis výsledku anglicky

In this paper we examine multiple methods for solving the problem of text vectorization in context of text classification. We compare two variants of a traditional Bag-of-Words technique to a newly proposed Word-Graph approach based on graph representation of a text document and measuring similarities between graph structures. We further propose modifications to the Word-Graph method potentially improving classification accuracy. Results of experiments performed while solving an author identification problem on a dataset consisting of speeches made during meetings of Slovak National Parliament show that the Word-Graph approach offers similar levels of accuracy as traditional methods. Proposed modifications significantly improve the performance in case of imbalanced number of documents for each class in the training set. (C) 2019 VSB-Technical University of Ostrava. All rights reserved.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2019

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of the 13th International Conference on Strategic Management and its Support by Information Systems: May 21th-22th, 2019, Ostrava, Czech Republic

ISBN

978-80-248-4305-6

ISSN

2570-5776

e-ISSN

—

Počet stran výsledku

8

Strana od-do

418-425

Název nakladatele

VŠB - Technical University of Ostrava

Místo vydání

Ostrava

Místo konání akce

Ostrava

Datum konání akce

21. 5. 2019

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—