Citation Number
2
Views
11
Downloands
3
Günümüzde gittikçe yaygınlaşan sosyal medya kullanımı ile duygular ve fikirler bu platformlar üzerinden ifade edilmektedir. Bu platformlarda paylaşılan fikirler ile büyük miktarda veri ortaya çıkmaktadır. Bu verilerin sınıflandırılmasının ve analizinin manuel olarak yapılması büyük bir iş gücü gerektirdiğinden bazı algoritmalar ile duygu analizi yapılması gereksinimi ortaya çıkmıştır. Bu çalışmada çeşitli platformlardan alınan beş farklı veri kümesi ve her bir veri kümesi için dört farklı makine öğrenmesi algoritması(KNN, Naif Bayes, Rastgele Orman, DVM) kullanılmıştır. Çalışma sonucunda DVM algoritması ile veri setlerinin genelinde daha doğru sonuçlar, Rastgele Orman ve Naif Bayes algoritmaları ile veri setleri ve eğitim yüzdelerine göre değişken sonuçlar elde edilmiştir. KNN algoritması ile veri setlerinin genelinde doğruluğu en düşük sonuçlar elde edilmiştir.
Today, with the increasingly widespread use of social media, emotions and ideas are expressed through these platforms. A large amount of data appears with the ideas shared on these platforms. As the classification and analysis of these data manually requires a great workforce, it has emerged the need to do emotional analysis with some algorithms. The study used five different datasets obtained from different platforms and four different machine learning algorithms (KNN, Naif Bayes, Rastgele Forest, DVM) for each datasets. The study resulted in more accurate results across the data sets with the DVM algorithm, random Forest and Naif Bayes algorithms and variable results according to data sets and training percentages. The KNN algorithm has achieved the lowest results of accuracy throughout data sets.
Recently, with the use of social media, which is becoming more and more widespread today, emotions and ideas are expressed through these platforms. Huge amounts of data emerge with ideas shared on these platforms. Since the classification and analysis of these data requires a large labor force, the need for sentiment analysis with some algorithms has emerged. In this study, five different datasets from various platforms and four different machine learning algorithms (kNN, Naive Bayes, Random Forest, SVM) were used for each dataset. As a result of the study, more accurate results were obtained in general with the SVM algorithm, and variable results were obtained with the Random Forest and Naive Bayes algorithms according to the data sets and training percentages. With the KNN algorithm, the lowest accuracy results were obtained across the data sets.
Field : Fen Bilimleri ve Matematik; Mühendislik
Journal Type : Uluslararası
Article Detail 
Citation Owners
Similar Articles
Download PDF
Journal Information
Listen Paper
Print This Page
Share
Mendeley
