January 04, 2019
Faculty Research: INFORMS Spotlight on Param Vir SinghINFORMS, The Institute for Operations Research and the Management Sciences, is the leading international association for professionals in operations research and analytics. Param Vir Singh, Carnegie Bosch Associate Professor of Business Technologies and INFORMS member since 2008, was interviewed about his research on "Copycats vs. Original Mobile Apps: A Machine Learning Copycat-Detection Method and Empirical Analysis," published in Information Systems Research.
INFORMS: What inspired you to research this particular topic?
SINGH: A few years ago, my coauthors, Quan Wang and Beibei Li, recommended me to try a mobile app game called Candy Crush. They thought it was too much fun. I had also heard about it, so I proceeded to download it from iTunes but got confused. There were just too many games that appeared to be the Candy Crush game. I could not figure out which one I should download. When I discussed this with Quan and Beibei, we all were confused as to which one is the original. This acted as a seed for our paper where we thought if we were having a problem identifying the original, so would other users. This could be bad for the developers who create original apps. In fact, when we searched online, we found numerous media and blog reports where practitioners and developers of original apps claimed that copycats steal the original app’s idea and potential demand, and were calling for app platforms to take action against such copycats. Surprisingly, however, we also found that there was little rigorous research analyzing whether and how copycats affect an original app’s demand. The primary deterrent to such research was the lack of an objective way to identify whether an app is a copycat or an original. This is when we decided to develop a method to identify which app is the original and which is a copycat and how a copycat affects the demand of the original.
INFORMS: Did any of your results surprise you?
SINGH: The result that surprised us was that, despite all the claims made by the developers of original apps, in aggregate, the effect of copycats on the demand of the original app was statistically insignificant.
INFORMS: What is the most important takeaway you hope readers will learn from your paper?
SINGH: The key takeaway is the machine learning-based method that we developed, which is scalable and easily implementable to determine original versus copycat apps. This opens up research opportunities for others to answer related questions.
The copycats in aggregate are not actually bad for the original apps. While the high-quality nondeceptive copycats take demand away from the original, the low-quality deceptive copycats drive demand to the original.
INFORMS: Briefly describe the notion of deceptive and nondeceptive copycat apps.
SINGH: Copycats are apps that copy the features and gameplay of the original apps. Deceptive copycats try to deceive the user by choosing a name as well as a logo that are deceptively similar to that of the original app. Nondeceptive copycats try to differentiate by choosing a name or logo that looks different from those of the original app.
INFORMS: Tell us about the process of writing this paper.
SINGH: We collected detailed panel data on thousands of mobile apps for the iOS platform. Our first challenge was to determine a way to identify which is an original and which is a copycat. This took quite some time. We needed to identify a quantifiable strategy to say two apps are the same. We determined the similarity based on the game play for the apps, which we determined based on the textual description and the customer reviews for the apps. We used natural language processing techniques to extract features of the app from the textual description and customer reviews. We then deployed a clustering approach to cluster apps that were similar in game play. We now needed to identify which of these apps were copycats and which was the original. We did this by using the release date of the first version of the app. The app that released the earliest was classified as the original app in the cluster. To identify deceptiveness we employed image analytics on the logos of the apps and text analytics on the names of the apps. Once the apps were identified as original or copycats (deceptive or nondeceptive), we performed the econometric analysis for identifying the impact of the copycat on the original.
INFORMS: Why was it important for you to publish in Information Systems Research?
SINGH: Information Systems Research (ISR) is the top journal in the field of information systems, and is both quantitatively rigorous and appreciates practicality. We felt both the reviewers as well as the readers of ISR would be the right audience for this work. This is one of the few journals where the reviewers are well-versed in both machine learning methods and econometrics. Their constructive comments helped us improve the paper.
INFORMS: How do you define “analytics”?
SINGH: In my view, analytics is using any method (computer science, economics, operations research, statistics, etc.) to discover, interpret, and communicate useful patterns in data.
INFORMS: How do you keep yourself up-to-date on the latest research in your field?
SINGH: Attending INFORMS conferences and reading journals are really important for keeping up-to-date on the latest research in the field. The doctoral students whom I work with at Carnegie Mellon are another tremendous resource for me to get introduced to new ideas. Increasingly, I believe the boundaries between areas are disappearing. Because Carnegie Mellon is an interdisciplinary institution, I regularly get exposed to ideas from across areas, which results in new ways to look at the same question.
INFORMS: What about your career might surprise us?
SINGH: Recently, I am leading a CMU-wide effort (with Ramayya Krishnan, Dean of Heinz College at Carnegie Mellon and INFORMS President-Elect) to develop a cryptocurrency/blockchain test bed (CMU Coin) at Carnegie Mellon for educational and research purposes. I am quite excited about this opportunity.
INFORMS: Are you currently doing any research? If so, can you tell us a little bit about what you’re working on?
SINGH: I am working on identifying the economic value of images in the context of Airbnb. We are identifying how a property image posted by a host affects demand. The overall objective is to understand what makes a good property image for an Airbnb property. We are combining deep learning techniques with causal analysis methods to answer this question. We have found some very good results that have tremendous practical value.
INFORMS: You were one of the first recipients of the Information Systems Society Sandra A. Slaughter Early Career Award. Tell us how this has affected your career.
SINGH: Sandra Slaughter is a huge inspiration for me. I feel even more close to her as she was a faculty member at Carnegie Mellon. Sandy was a department editor for Management Science when I had just graduated. As editor and a senior colleague in the field, she played an important and constructive role in shaping my research. The Sandra A. Slaughter Early Career Award is the most prestigious award in the field of information systems for young faculty. I am very honored as well as humbled to be one of the first recipients of the award. I strive to live up to the standards that Sandy would have expected.
INFORMS: When you’re not using your OR/MS superpowers to try to make the world a better place, what are some of the ways you like to spend your time?
SINGH: I like to learn and explore new things, particularly related to technology.
INFORMS: Which INFORMS event do you most look forward to attending?
SINGH: I have been attending the INFORMS Annual Meeting and Conference on Information Systems and Technology (CIST) for the last 10+ years and will continue to attend in the future as well. Both of these conferences are excellent events to meet with researchers from across disciplines working on cutting-edge research.
INFORMS: How do you relax?
SINGH: Play with my kids Elin and AD.
INFORMS: Which social network do you use most and why?
SINGH: Facebook, WhatsApp, WeChat, and LinkedIn. All have become key sources of information.
INFORMS: What is the best advice you can give to students in your field?
SINGH: Ph.D. is the time when a student should “aim big and dig deep & wide.” By “aim big,” I mean go for big ideas — something that would be fundamentally new. Students should go out and be exposed to research happening in other fields including computer science, economics, and other business disciplines. When you combine ideas from across areas, you have a real opportunity to create something fundamentally new. At the same time, students need to be careful that they do not end up being mediocre in multiple areas. Their goal should be to “dig deep and wide” where they become experts in multiple areas. In the past, it was not possible, but with technology, it is increasingly possible as all the knowledge is easily available online.