Toba, Hapnes and Stefanus, William (2016) Prediksi Profesi Berdasarkan Model Bahasa pada Tweets. In: Seminar Nasional Teknologi Informasi dan Multimedia 2016, 6-7 Februari 2016, Yogyakarta.
|
Text
22.pdf - Published Version Download (1616Kb) | Preview |
Abstract
With the advance of social media, people tends to be very reactive on issues which are happening around the globe. Everybody can show their opinions freely, and sometimes uncontrollable, no matters what their job is. This research investigates the tendency of words choice in someone’s job based on the style of language he/she used in his/her twitter account. It is assume that most of the people in a specific job has the same language used on social media. The analyses of the study is performed by using Naïve Bayes classifiers for around 30,000 tweets. The text processing are divided into three main parts, i.e.: retrieval and grouping of the data, data processing, and evaluation. The type of jobs which are analyzed, consists of: politicians, actresses/actors, musicians, and students, through their official twitter accounts. The experimental results show that multinomial Bayes classifiers are more reliable than the binomial classifiers. Further investigation shows that the best accuracy is achieved by the unigram model, which has a mean of 0.73±0.127 in a 5 cross validation setting. This fact reveals that there is no direct relatioship between someone’s word choice and his/her profession.
| Item Type: | Conference or Workshop Item (Paper) |
|---|---|
| Uncontrolled Keywords: | text analysis, language (n-grams) models, job prediction, Naïve Nayes classification, social media content analysis. |
| Subjects: | T Technology > T Technology (General) |
| Depositing User: | Perpustakaan Maranatha |
| Date Deposited: | 26 Sep 2018 07:20 |
| Last Modified: | 10 Oct 2018 10:32 |
| URI: | http://repository.maranatha.edu/id/eprint/24787 |
Actions (login required)
 |
View Item |
