Klasifikasi Sentimen Menggunakan Semi-supervised Subjective Feature Weighting and Intelligent Model Selection pada Forum Diskusi Online

Bagaskarta, Adam Widi (2018) Klasifikasi Sentimen Menggunakan Semi-supervised Subjective Feature Weighting and Intelligent Model Selection pada Forum Diskusi Online. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
05111440000089-Undergraduate_Theses.pdf - Published Version
Restricted to Repository staff only

Download (2MB) | Request a copy

Abstract

Jumlah pengguna internet di seluruh dunia bertambah dengan pesat dari 2 miliar di tahun 2010 sekarang menjadi 3 miliar pengguna seluruh dunia ditahun 2017. Bertambahnya penggunaan akses internet yang mudah, membuat sosial media menjadi sangat populer di kalangan masyarakat. Masyarakat akan cenderung bersosialisasi dengan cara mengekspresikan opini melalui sosial media yang ada seperti facebook, twitter, instagram, dan lain-lain. Dalam tugas akhir ini, akan dilakukan klasifikasi sentimen menggunakan data yang berasal dari twitter. Data tersebut meliputi tweet dari akun XL, Telkomsel, dan IM3. Dengan mengetahui hasil sentimen pada sebuah produk, pihak produsen akan dapat merancang sebuah rencana bisnis untuk meningkatkan pelayanan kepada konsumen. Data tersebut akan di klasfikasi menggunakan metode Semi-supervised Subjective Feature Weighting and Intelligent Model Selection (SWIMS). Untuk hasil klasifikasi data tersebut akan dilakukan perbadingan antara penggunaan jumlah fitur menggunakan POS Tagging, Term Presence (TP), dan gabungan dari kedua fitur tersebut. Dari hasil tersebut didapatkan bahwa penggunaan gabungan antara fitur POS Tagging dan Term Presence (TP) serta menggunakan kernel linear pada SVM memberikan hasil terbaik yaitu accuracy sebesar 82,5%, precision 82,7%, recall 83,5%, dan f-measure 82,4%. ============ The number of internet users worldwide increased rapidly from 2 billion in 2010 now to 3 billion users worldwide in 2017. Increasing use of easy Internet access, making social media become very popular among the public. The community will tend to socialize by way of expressing opinions through existing social media such as facebook, twitter, instagram, and others. In this study, will be classified sentiment using data derived from twitter. The data includes tweets from XL, Telkomsel and IM3 accounts. By knowing the sentiment of a product, the manufacturer will be able to design a business plan to improve service to consumers. The data will be classified using the Semi-supervised Bubjective Feature Weighting and Intelligent Model Selection (SWIMS) method. For the results of the data classification will be made a comparison between the use of the number of features using POS Tagging, Term Presence (TP), and a combination of both features. From these result, it was found that the combined use of both features which is POS Tagging and Term Presence (TP) gave the best result of accuracy of 82.5%, 82.7% precision, 83.5% recall, and f-measure 82.4%.

Item Type: Thesis (Undergraduate)
Additional Information: RSIf 006.7 Bag k
Uncontrolled Keywords: twitter; SWIMS; SVM; cross validation; sentiwordnet
Subjects: Q Science > QA Mathematics > QA336 Artificial Intelligence
Q Science > QA Mathematics > QA76.9.D343 Data mining
T Technology > T Technology (General) > T57.5 Data Processing
Divisions: Faculty of Information and Communication Technology > Informatics > (S1) Undergraduate Theses
Depositing User: Adam Widi Bagaskarta
Date Deposited: 23 Jul 2018 07:31
Last Modified: 22 Nov 2018 20:18
URI: http://repository.its.ac.id/id/eprint/53009

Actions (login required)

View Item View Item