DETEKSI AKTIVITAS DENGAN MEMANFAATKAN DATA LAYANAN JEJARING SOSIAL BERBASIS LOKASI TWITTER

Futra, Muhammad Haryadi (2014) DETEKSI AKTIVITAS DENGAN MEMANFAATKAN DATA LAYANAN JEJARING SOSIAL BERBASIS LOKASI TWITTER. S1 thesis, Universitas Pendidikan Indonesia.

Abstract

Twitter merupakan salah satu layanan jejaring sosial berbasis lokasi yang kini sangat populer di kalangan masyarakat pengguna internet. Twitter menjadi sumber data yang sangat bermanfaat sekaligus menjadi salah satu pusat penyedia informasi yang bersifat real-time. Pada jurnal ini penulis menitikberatkan penelitian pada pendeteksian aktivitas dengan memanfaatkan kumpulan tweet para pengguna Twitter. Untuk melakukan penelitian ini digunakan metode-metode seperti metode klasifikasi dengan algoritma Naive Bayes dan metode clustering dengan algoritma K-Means. Metode klasifikasi digunakan untuk memisahkan kumpulan tweet ¬¬ke dalam kelas-kelas yang telah ditentukan, dan metode clustering digunakan untuk mengelompokkan kumpulan tweet yang telah diklasifikasi ke dalam cluster-nya masing-masing berdasarkan informasi aktivitas yang terdapat di dalamnya, dan setiap cluster mewakili satu aktivitas. Deteksi aktivitas dengan metode klasifikasi dan clustering memberikan hasil yang terbilang baik, dibuktikan dengan nilai F-Measure yang diperoleh untuk metode klasifikasi dengan algoritma Naive Bayes yaitu sebesar 77,068 %, dan nilai purity untuk metode clusering dengan algoritma K-Means yaitu sebesar 0,599.
Kata Kunci: Twitter is a location-based social networking service that is now very popular among netizens. Twitter is a very useful source of data and become one of the central provider of real time information. In this paper the authors study focuses on detecting activity by utilizing a collection of Twitter users tweet. To conduct this study, used methods such as the method of classification with Naive Bayes algorithm and method of clustering with K-Means algorithm. Classification method is used to separate the tweet into classes that have been determined, and the clustering method is used to classify tweets that have been classified into each cluster based on the information about the activities contained in it. Each cluster represents a single event. Activity detection with classification and clustering methods give fairly good results, evidenced by the value obtained for the F-Measure of the naive Bayes classification method of 77.068%, and the value of purity for clusering method with K-Means algorithm by 0.599.
Keywords: Activity, Twitter, Tweet, Classification, Clustering, Naive Bayes, ¬K-Means, Purity.

Twitter is a location-based social networking service that is now very popular among netizens. Twitter is a very useful source of data and become one of the central provider of real time information. In this paper the authors study focuses on detecting activity by utilizing a collection of Twitter users tweet. To conduct this study, used methods such as the method of classification with Naive Bayes algorithm and method of clustering with K-Means algorithm. Classification method is used to separate the tweet into classes that have been determined, and the clustering method is used to classify tweets that have been classified into each cluster based on the information about the activities contained in it. Each cluster represents a single event. Activity detection with classification and clustering methods give fairly good results, evidenced by the value obtained for the F-Measure of the naive Bayes classification method of 77.068%, and the value of purity for clusering method with K-Means algorithm by 0.599.
Keywords: Activity, Twitter, Tweet, Classification, Clustering, Naive Bayes, ¬K-Means, Purity.

Preview	Text S_KOM_1002484_Title.pdf Download (11kB) | Preview
Preview	Text S_KOM_1002484_Abstract.pdf Download (134kB) | Preview
Preview	Text S_KOM_1002484_Table_ Of_ Content.pdf Download (226kB) | Preview
Preview	Text S_KOM_1002484_Chapter1.pdf Download (145kB) | Preview
	Text S_KOM_1002484_Chapter2.pdf Restricted to Staf Perpustakaan Download (479kB)
Preview	Text S_KOM_1002484_Chapter3.pdf Download (117kB) | Preview
	Text S_KOM_1002484_Chapterr4.pdf Restricted to Staf Perpustakaan Download (581kB)
Preview	Text S_KOM_1002484_Chapter5.pdf Download (134kB) | Preview
Preview	Text S_KOM_1002484_Bibliography.pdf Download (207kB) | Preview
	Text S_KOM_1002484_Appendix.pdf Restricted to Staf Perpustakaan Download (378kB)

Item Type:	Thesis (S1)
Additional Information:	No.Panggil: S KOM FUT -(2014)
Uncontrolled Keywords:	Twitter is a location-based social networking service that is now very popular among netizens. Twitter is a very useful source of data and become one of the central provider of real time information. In this paper the authors study focuses on detecting activity by utilizing a collection of Twitter users tweet. To conduct this study, used methods such as the method of classification with Naive Bayes algorithm and method of clustering with K-Means algorithm. Classification method is used to separate the tweet into classes that have been determined, and the clustering method is used to classify tweets that have been classified into each cluster based on the information about the activities contained in it. Each cluster represents a single event. Activity detection with classification and clustering methods give fairly good results, evidenced by the value obtained for the F-Measure of the naive Bayes classification method of 77.068%, and the value of purity for clusering method with K-Means algorithm by 0.599. Keywords: Activity, Twitter, Tweet, Classification, Clustering, Naive Bayes, ¬K-Means, Purity.
Subjects:	Q Science > Q Science (General)
Divisions:	Fakultas Pendidikan Matematika dan Ilmu Pengetahuan Alam > Program Studi Ilmu Komputer
Depositing User:	DAM staf
Date Deposited:	16 Feb 2015 02:49
Last Modified:	16 Feb 2015 02:49
URI:	http://repository.upi.edu/id/eprint/13050

Actions (login required)

View Item