Home // AFIN 2014, The Sixth International Conference on Advances in Future Internet // View article

Sentiment Analysis on Online Social Network Using Probability Model

Authors:
Hyeoncheol Lee
Youngsub Han
Kwangmi Kim

Keywords: sentiment analysis, social network

Abstract:
Sentiment analysis is to extract people¡¯s opinion and knowledge from text messages. Recently, demands on automated sentiment analysis tool for text messages generated from web have dramatically increased and the literature on this topic has been growing. In this paper, we propose a semi-automated sentiment analysis method on online social network using probability model. The proposed method reads sample text messages in a train set and builds a sentiment lexicon that contains the list of words that appeared in the text messages and probability that a text message is positive opinion if it includes those words. Then, it computes the positivity score of text messages in a test set using the list of words in a message and sentiment lexicon. Each message is categorized as either positive or negative, depending on threshold value calculated using a train set. To check the accuracy, we compared the sentiments of the proposed method with sentiments of human coders. This research is unique and novel in that it guarantees high accuracy rates and does not require additional information, such as users¡¯ profile and network relationship.

Pages: 14 to 19

Copyright: Copyright (c) IARIA, 2014

Publication date: November 16, 2014

Published in: conference

ISSN: 2308-4340

ISBN: 978-1-61208-377-3

Location: Lisbon, Portugal

Dates: from November 16, 2014 to November 20, 2014