App developers live and die by reviews. Lots of one-star ratings can sink an app quickly. But figuring out why reviewers aren’t happy hasn’t always been easy.
Now, a team of researchers has up with a new way to mine the text to make sense of the reviews in one step.
The idea was, can you devise a method that would look through all the ratings, and say these are the topics people are unhappy about and this is maybe where a developer should focus,” said Shawn Mankad, assistant professor of operations, technology and information management in Cornell’s Samuel Curtis Johnson Graduate School of Management, in a statement about the new technology.
Mankad is lead author of “Single Stage Prediction with Embedded Topic Modeling of Online Reviews for Mobile App Management,” which is to be published in the Annals of Applied Statistics. Co-authors of the study are Cornell doctoral candidate Shengli Hu and Anandasivam Gopal of the University of Maryland.
One way to mine text is to build a matrix to keep track of words that appear in reviews. “It becomes a really wide matrix. And you have so many columns that you need to shrink them down somehow,” Mankad said. “So that’s where we’re applying the method.”
The method creates dashboards that can help developers know how an app is doing. It also can compare it to competing apps over time.
The researchers determined that their text-mining model works better at forecasting accuracy on real reviews and simulated data than standard methods. Use of the new model can help developers decide when to release new versions of apps.