13 Abr Today, there are lots of relationship applications that are commonly used, such as the well-known Tinder and Okcupid
Considering the expanding interest in relationships software additionally the unsatisfactory affiliate evaluations off biggest relationships applications, we made a decision to learn an individual recommendations from relationship software having fun with one or two text mining measures. Earliest, i built an interest model based on LDA so you can mine the latest bad evaluations away from conventional relationships apps, analyzed part of the reason why users render negative ratings, and place forward corresponding update recommendations. Second, i based a-two-stage machine reading model one joint data dimensionality protection and you may study class, hoping to get a definition that can effectively identify user reviews regarding relationships programs, so that software providers can also be techniques reading user reviews better.
dos.step one Investigation order
Since most pages install these types of software of Bing Play, i thought that software product reviews on google Play is also effortlessly mirror representative feelings and perceptions into the this type of software. Most of the study we used are from recommendations of profiles from these six relationship apps: Bumble, Java Matches Bagel, Depend, Okcupid, Enough Seafood and you may Tinder. The information is wrote to your figshare , i vow you to sharing this new dataset on the Figshare complies into the terms and conditions of your internet at which data is actually utilized. Together with, we pledge your ways of analysis range put and its application within our studies conform to the new regards to this site from which the details started. The data are the text message of one's recommendations, what amount of likes user reviews get, as well as the reviews' evaluations of one's programs. At the end of , i've accumulated all in all, 1,270,951 reviews analysis.