Analisis Sentimen untuk Komentar pada Sistem Pencarian Kost Menggunakan Metode Support Vector Machine (SVM)

Abstract

With limited information for college students who want to find boarding, effort and time needed to obtain information. Then boarding search sistem will be needed to facilitate in finding a boarding. On this basis the proposed making boarding search system with the rating system information and recommendations boarding house around SWCU which will facilitate in finding a boarding. For the rating system on the final assignment was filed with sentiment analysis system using the method of Support Vector Machine (SVM). Then on the web system, there are 3 types of users namely Admin, Owner of Boarding and College Students. Admin duty to check the authenticity of the data for each boarding and can manage all of the existing web activity. Then the boarding house owner can promote a boarding house with input data through the approval of the admins. And then the college students can searching with a web for the boarding house that has been filled, then if the college student is already in the boarding house, they can do the review with comments and stars. Then SVM here that would classify classify comments – comments on each boarding into the shape of positive and negative values. So it can be used to calculate the value of the rating on a scale of 0 – 5. The rating of it will pop up the recommendations boarding with the highest rating will appear on the main page of the web. Tests conducted on five different boarding place with each boarding house has at least 11 comments will be analyzed by SVM to determine including positive or negative sentences. Testing of SVM in five different boarding places using parameter C = 2.5 and  = 0.3 obtained at boarding places Turen II received 92.30% accuracy, Dipo 88 received 81.81% accuracy, Kemiri 2 received 92.30%, Graha Widya received 54.54% and Wisma Mawar received 90.90% accuracy.