사이버 공론장의 남성성과 여성성 판별

In this research, we report experimental results on our attempt to detect gender from debates in Korean cyber public sphere using machine learning techniques for text classification. For the experiments, we have collected for a certain periods posted articles from two online cyber public spheres, each of which only male or female can post their opinions. From the collected articles, we construct naive Bayes classifiers with multinomial model. Using the constructed classifiers, we perform classification on a newly posted articles to analyze the their performance. Finally, we analyze the degree of genderness of each term by its likelihood ratio from measured parameters. We anticipate the experimental results on the cyber public sphere debate data can contribute future research areas about mining public opinions, etc.