Sentiment document classification using global and domain features

Youngjoong Ko

Sentiment document classification using global and domain features

Youngjoong Ko

Dong-A University

Research output: Contribution to journal › Article › peer-review

Abstract

The goal of sentiment classification is to detect writer's sentiment from a document. This paper investigates which features and what combination of them is more effective in sentiment classification. Experiments show that the effective combination method of global and domain features can significantly reduce classification errors relative to features which have been used in general text classification.

Original language	English
Pages (from-to)	699-707
Number of pages	9
Journal	Information
Volume	17
Issue number	2
State	Published - Feb 2014
Externally published	Yes

Keywords

Feature selection
Opinion mining
Sentiment classification
Sentiment-bearing words

Cite this

@article{086339dbc28a40b2bd9db80daa1caef4,

title = "Sentiment document classification using global and domain features",

abstract = "The goal of sentiment classification is to detect writer's sentiment from a document. This paper investigates which features and what combination of them is more effective in sentiment classification. Experiments show that the effective combination method of global and domain features can significantly reduce classification errors relative to features which have been used in general text classification.",

keywords = "Feature selection, Opinion mining, Sentiment classification, Sentiment-bearing words",

author = "Youngjoong Ko",

year = "2014",

month = feb,

language = "English",

volume = "17",

pages = "699--707",

journal = "Information",

issn = "1343-4500",

publisher = "International Information Institute",

number = "2",

}

TY - JOUR

T1 - Sentiment document classification using global and domain features

AU - Ko, Youngjoong

PY - 2014/2

Y1 - 2014/2

N2 - The goal of sentiment classification is to detect writer's sentiment from a document. This paper investigates which features and what combination of them is more effective in sentiment classification. Experiments show that the effective combination method of global and domain features can significantly reduce classification errors relative to features which have been used in general text classification.

AB - The goal of sentiment classification is to detect writer's sentiment from a document. This paper investigates which features and what combination of them is more effective in sentiment classification. Experiments show that the effective combination method of global and domain features can significantly reduce classification errors relative to features which have been used in general text classification.

KW - Feature selection

KW - Opinion mining

KW - Sentiment classification

KW - Sentiment-bearing words

UR - https://www.scopus.com/pages/publications/84900014942

M3 - Article

AN - SCOPUS:84900014942

SN - 1343-4500

VL - 17

SP - 699

EP - 707

JO - Information

JF - Information

IS - 2

ER -

Sentiment document classification using global and domain features

Abstract

Keywords

Other files and links

Fingerprint

Cite this