Abstract
To the continuing growth of it, more individuals have become to utilize dating software, and the pattern has been exacerbated because of the COVID-19 pandemic during these years. Although not, there is certainly a sensation that all of reading user reviews from main-stream relationship programs is actually negative. To study so it experience, you will find made use of thing design to help you mine negative ratings regarding main-stream dating programs, and you may built a two-stage host learning model playing with study dimensionality reduction and text message category so you can classify user reviews out of dating applications. The study overall performance show that: to start with, the reasons toward newest negative feedback of dating programs is actually primarily centered on recharging system, phony account, membership and advertisements force system and complimentary apparatus regarding the apps, advised related improve information was suggested from the all of us; furthermore, playing with principal component study to attenuate the fresh dimensionality of your text message vector, right after which having fun with XGBoost model to learn the reduced-dimensional investigation after oversampling, a better category accuracy off user reviews can be obtained. Continue reading