• Journal of the Korean Institute of Intelligent Systems
• /
• v.24 no.5
• /
• pp.482-488
• /
• 2014

This paper aims to analyze user's emotion automatically by analyzing Twitter, a representative social network service (SNS). In order to create sentiment analysis models by using machine learning techniques, sentiment labels that represent positive/negative emotions are required. However it is very expensive to obtain sentiment labels of tweets. So, in this paper, we propose a sentiment analysis model by using self-training technique in order to utilize "data without sentiment labels" as well as "data with sentiment labels". Self-training technique is that labels of "data without sentiment labels" is determined by utilizing "data with sentiment labels", and then updates models using together with "data with sentiment labels" and newly labeled data. This technique improves the sentiment analysis performance gradually. However, it has a problem that misclassifications of unlabeled data in an early stage affect the model updating through the whole learning process because labels of unlabeled data never changes once those are determined. Thus, labels of "data without sentiment labels" needs to be carefully determined. In this paper, in order to get high performance using self-training technique, we propose 3 policies for updating "data with sentiment labels" and conduct a comparative analysis. The first policy is to select data of which confidence is higher than a given threshold among newly labeled data. The second policy is to choose the same number of the positive and negative data in the newly labeled data in order to avoid the imbalanced class learning problem. The third policy is to choose newly labeled data less than a given maximum number in order to avoid the updates of large amount of data at a time for gradual model updates. Experiments are conducted using Stanford data set and the data set is classified into positive and negative. As a result, the learned model has a high performance than the learned models by using "data with sentiment labels" only and the self-training with a regular model update policy.

• The Journal of Society for e-Business Studies
• /
• v.20 no.3
• /
• pp.59-77
• /
• 2015

Social networking sites (SNS), such as Facebook, provide abundant social comparison opportunities. Given the widespread use of SNSs, the purpose of the present study was to examine the impact of exposure to social media-based social comparison on user's negative emotions and discontinuous use intention on SNS. We present evidence that under the use of SNS, social comparison activities diverge into three patterns, with explicit self-evaluation desire made against similar target (lateral comparison), self-defense desire made against less fortunate target (downward comparison), and self-enhancement desire made with more fortunate target (upward comparison). Such social comparison processes frequently arise, as people are increasingly using on SNSs, the downward contacts ameliorating self-esteem with positive emotions, but the upward contacts and standard contacts with lateral status enabling a person to compare his or her situation with others and simultaneously increase negative emotions due to its differences with others. In other words, as people increasingly relying on SNSs for a variety of everyday tasks, they risk overexposure to upward or standard social comparison information that may have a cumulative detrimental impact on future intention on SNS use. This study with survey with 209 SNS users found that these negative emotions lead to negative fatigue (attitude) and then discontinuous use intention (behavior) on SNS. Our findings are among the first to explicitly examine discontinuous use intention on SNS using social comparison theory and our results are consistent with those of past research showing that upward social comparisons can be detrimental.

• Journal of Korea Entertainment Industry Association
• /
• v.14 no.6
• /
• pp.139-150
• /
• 2020

The data processing of this study focuses on the textom and social media words about three areas: 'Corona 19 and professional baseball', 'Corona 19 and professional baseball', and 'Corona 19 and professional sports' The data was collected and refined in a web environment and then processed in batch, and the Ucinet6 program was used to visualize it. Specifically, the web environment was collected using Naver, Daum, and Google's channels, and was summarized into 30 words through expert meetings among the extracted words and used in the final study. 30 extracted words were visualized through a matrix, and a CONCOR analysis was performed to identify clusters of similarity and commonality of words. As a result of analysis, the clusters related to Corona 19 and Pro Baseball were composed of one central cluster and five peripheral clusters, and it was found that the contents related to the opening of professional baseball according to the corona 19 wave were mainly searched. The cluster related to Corona 19 and unrelated to professional baseball consisted of one central cluster and five peripheral clusters, and it was found that the keyword of the position of professional baseball related to the professional baseball game according to Corona 19 was mainly searched. Corona 19 and the cluster related to professional sports consisted of one central cluster and five peripheral clusters, and it was found that the keywords related to the start of professional sports according to the aftermath of Corona 19 were mainly searched.