Mobile app developers need to react quickly to reviews that reflect dissatisfaction. Statisticians at Cornell have found a faster way for developers to respond, with a new text-mining method that aggregates and parses customer reviews in one step.

The innovative model was devised by Shawn Mankad, assistant professor of operations, technology and information management in the Samuel Curtis Johnson Graduate School of Management. With his team, Mankad was, in 2017, awarded a grant of $525,000 over 4 years to create new tools to monitor the stability of the financial system. His paper, Single stage prediction with embedded topic modeling of online reviews for mobile app management, co-authored with Cornell doctoral candidate Shengli Hu and Anandasivam Gopal of the University of Maryland has been accepted for the Annals of Applied Statistics.

In the paper's abstract Mankad et al. state:

A diferentiating feature of mobile apps to traditional
enterprise software is online reviews, which are available on app marketplaces and represent a valuable source of consumer feedback on the app. We create a supervised topic modeling approach for app developers to use mobile reviews as useful sources of quality and customer feedback, thereby complementing traditional software testing. The approach is based on a constrained matrix factorization that leverages the relationship between term frequency and a given response variable in addition to co-occurrences between terms to recover topics that are both predictive of consumer sentiment and useful for understanding the underlying textual themes. 

Explaining the method in more informal terms. Mankad notes that in text mining, a common way to represent texts is to construct a huge matrix to keep track of which words appear in which online review. The matrix becomes very wide and it is necessary to reduce the number of columns. The model, in effect, takes a weighted average of words that appear in online reviews. Each of those weighted averages represents a topic of discussion. The method not only provides guidance on a single app's performance but also compares it to competing apps over time to benchmark features and consumer sentiment.

Credit: Cornell Brand Communications

Mankad and his colleagues applied their approach to both simulated data and more than 104,000 mobile reviews of 162 versions of apps from three of the most popular online travel agents in the United States: Expedia, Kayak and TripAdvisor. There were more than 1,000 reviews per app per year. They found that their text-mining model performed better than the standard methods at forecasting accuracy on both real reviews and simulated data. And they found that the method can help companies weigh the pros and cons of how frequently they release new versions of their apps.

More Information

Single stage prediction with embedded topic modeling of online reviews for mobile app management

Thanks, statistics! A faster way to improve mobile apps

Related Articles

Why Users Uninstall Apps

Business Apps - An Untapped Developer Opportunity

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info