Study on Methods for Vector Representation of Text for Topic-based Clustering of News Articles

Kód výsledku v IS VaVaI

RIV/46747885:24220/15:#0003432 - isvavai.cz

Nalezeny alternativní kódy

RIV/46747885:24220/15:00002977

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Study on Methods for Vector Representation of Text for Topic-based Clustering of News Articles

Popis výsledku v původním jazyce

This paper deals with methods for automatic topic-based clustering of news articles. For this purpose, three approaches are evaluated experimentally using two different test sets that are both compiled from Czech newspaper articles. The first adopted clustering scheme utilizes conventional approach: Latent Semantic Analysis (LSA). In contrast, the remaining two approaches were introduced recently and utilize Random Manhattan Indexing (RMI) or Skip-gram model (SGM) to obtain vector representation of input words and/or documents. Our experimental results show that both the new vectorization methods are for topic-based clustering well suitable and that their lead to even better results than LSA. The best clustering accuracy was reached by SGM and it is by17 % lower than the accuracy of human annotators. On the other hand, RMI yielded just slightly worse results than SGM and has one important advantage for practical usage: it can handle out of vocabulary (OOV) words without the need for t

Název v anglickém jazyce

Study on Methods for Vector Representation of Text for Topic-based Clustering of News Articles

Popis výsledku anglicky

This paper deals with methods for automatic topic-based clustering of news articles. For this purpose, three approaches are evaluated experimentally using two different test sets that are both compiled from Czech newspaper articles. The first adopted clustering scheme utilizes conventional approach: Latent Semantic Analysis (LSA). In contrast, the remaining two approaches were introduced recently and utilize Random Manhattan Indexing (RMI) or Skip-gram model (SGM) to obtain vector representation of input words and/or documents. Our experimental results show that both the new vectorization methods are for topic-based clustering well suitable and that their lead to even better results than LSA. The best clustering accuracy was reached by SGM and it is by17 % lower than the accuracy of human annotators. On the other hand, RMI yielded just slightly worse results than SGM and has one important advantage for practical usage: it can handle out of vocabulary (OOV) words without the need for t

Druh

D - Stať ve sborníku

CEP obor

JC - Počítačový hardware a software

OECD FORD obor

—

Projekt

TA04010199: MULTILINMEDIA - Multilinguální platforma pro monitoring a analýzu multimédií

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)
I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2015

Kód důvěrnosti údajů

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

Název statě ve sborníku

7th Language & Technology Conference: Human Language Technologies as a Challenge for Computer Science and Linguistics

ISBN

978-83-932640-8-7

ISSN

—

e-ISSN

—

Počet stran výsledku

5

Strana od-do

530-534

Název nakladatele

Fundancja Uniwersytetu im. Adama Mickiewicza w Poznaniu

Místo vydání

Polsko

Místo konání akce

Polsko, Poznaň

Datum konání akce

1. 1. 2015

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—