ANALISIS SENTIMEN TERHADAP FACEBOOK MARKETPLACE MENGGUNAKAN METODE LEXICON BASED DAN SUPPORT VECTOR MACHINE

SENTIMENT ANALYSIS OF FACEBOOK MARKETPLACE USING LEXICON BASED AND SUPPORT VECTOR MACHINE METHODS

Pandemi Covid-19 berdampak pada banyak hal, salah satunya budaya jual beli masyarakat. Jual beli masyarakat yang tadinya dilakukan secara langsung berubah menjadi jual beli secara online karena mengikuti aturan pemerintah untuk membatasi interaksi secara langsung maupun berkerumun. Facebook Marketplace adalah tempat jual beli secara online yang dikhususkan untuk pengguna Facebook karena berada didalam platform Facebook. Dibutuhkan data berupa opini mengenai Facebook Marketplace untuk melakukan analisis sentimen. Pengumpulan data opini diambil dari API Twitter karena Twitter merupakan media sosial yang banyak digunakan masyarakat untuk menampung opini dan paling update. Dari opini tersebut dapat dilihat sentimen tingkat kepuasan pengguna Facebook Marketplace dengan membaginya kedalam kelas sentimen negatif dan sentimen positif. Metode Lexicon Based digunakan untuk pembobotan opini sentimen positif dan negatif, kemudian proses klasifikasinya menggunakan metode Support Vector Machine (SVM). Hasil pembobotan menggunakan metode Lexicon Based menunjukkan hasil sentimen positif sebesar 82,6%, sedangkan sentimen negatif sebesar 17,4. Pada klasifikasi SVM menghasilkan nilai akurasi 51%, nilai presisi 51%, nilai recall 64%, dan AUC sebesar 0,51. pada kernel polynomial menghasilkan akurasi sebesar 52%, nilai presisi sebesar 84%, nilai recall sebesar 6%, dan AUC sebesar 0,52.

Kata Kunci— Pandemi Covid-19, Facebook Marketplace. Twitter, Lexicon based, Support Vector Machine (SVM).

The Covid-19 pandemic has had an impact on many things, one of which is the culture of buying and selling people. Community buying and selling which was previously carried out directly has turned into online buying and selling because it follows government regulations to limit direct or crowd interaction. Facebook Marketplace is an online buying and selling place that is specifically for Facebook users because it is on the Facebook platform. It takes data in the form of opinions about Facebook Marketplace to conduct sentiment analysis. Opinion data collection is taken from the Twitter API because Twitter is a social media that is widely used by the public to accommodate opinions and is the most updated. From this opinion, it can be seen that the sentiment of the satisfaction level of Facebook Marketplace users by dividing it into classes of negative sentiment and positive sentiment. The Lexicon Based method is used for weighting positive and negative sentiment opinions, then the classification process uses the Support Vector Machine (SVM) method. The results of weighting using the Lexicon Based method show positive sentiment results of 82.6%, while negative sentiment of 17.4%. The SVM classification produces an accuracy value of 51%, a precision value of 51%, a recall value of 64%, and an AUC of 0.51. the polynomial kernel produces an accuracy of 52%, a precision value of 84%, a recall value of 6%, and an AUC of 0.52. Keywords— Covid-19 pandemic, Facebook Marketplace. Twitter, Lexicon based, Support Vector Machine (SVM)