2016 INFORMS Annual Meeting Program

SD48

INFORMS Nashville – 2016

SD48

2 - Unlocking Your 80%: Unearthing New Insights With Text Analytics Christina Engelhardt, SAS Institute, Christina.Engelhardt@sas.com How can your organization harness the staggering volumes of textual data coming from social & online media and your own proprietary systems? Join us as we explore some of the challenges and exciting opportunities these rich, yet complex, data sources provide us. Session topics include: • Why you should consider incorporating text analytics into your data science, research, and operational work streams• How to align technology, data sources, and the various text methods with your use case and objectives• Examples of how leading organizations are leveraging Text Analytics 3 - Hierarchical Machine Learning Approach To Detecting Anomalous Behavior In Online Social Media Forums Naveen Kumar, University of Memphis, Memphis, TN, 38152, United States, nkumar7@memphis.edu, Deepak Venugopal, Robin Poston The detection of anomalous behavior in online social media is a challenging problem due to complex interactions between several user characteristics such as review veracity, velocity, volume, and variety. We propose a novel two stage hierarchical machine learning approach that increases the likelihood of detecting anomalies by analyzing different actions of individual users and then characterizing their collective behavior. Specifically, we model user characteristics as univariate/multivariate distributions and then combine these distributions using mixture models to obtain a unified view of a user’s behavior. We apply our approach to real-world reviews and obtain promising results. 212-MCC Dr. William Massey: A Dynamic Legacy Sponsored: Minority Issues Sponsored Session Chair: Jamol Pender, Cornell University, 206 Rhodes Hall, Ithaca, NY, 14853-3801, United States, jamol.pender@gmail.com 1 - Dynamic Rate Queues Jamol Pender, Cornell University, jamol.pender@gmail.com Inspired by healthcare and transportation systems, this talk will summarize the past, present, and future of dynamic rate queues and their impact on our society. 2 - Dr. William Massey: A Dynamic Legacy Robert Hampshire, University of Michigan, hamp@umich.edu In this talk we will explore the applications of time varying queues to problems in urban transportation. We show how Bill Massey’s fundamental contributions to queueing theory and applied probability can be applied to smart parking systems, bike sharing and car sharing services 3 - The Dynamics 0f Queueing Transience With Dynamic Rates William A Massey, Professor, Princeton University, ORFE Department, Sherrerd Hall, Princeton University, Princeton, NJ, 08544, United States, wmassey@princeton.edu Inspired by communication and healthcare services, this talk summarizes the methods developed with many collaborators over the decades to understand the transient behavior of dynamic rate queues. This analysis is needed when confronted with the dynamic parameters found in time-inhomogeneous Markovian queueing models. The static equilibrium analysis for the steady state of constant rate queues no longer applies. Constants summarizing the transient behavior for these steady state systems yield to the natural substitute of deterministic dynamical systems. We can then approximate the optimal behavior of these queues by controlling this related family of ordinary differential equations. SD50

210-MCC Social Media Analytics for Competitive Advantage Invited: Social Media Analytics Invited Session Chair: Vilma Todri, New York University, New York, NY, 11111, United States, vtodri@stern.nyu.edu 1 - Social Influence And Changing Circumstances In The Creation, Maintenance, And Disruption Of Habits In Global Health Behavior Christos Nicolaides, Massachusetts Institute of Technology, chrisnic@mit.edu In this research I analyze a unique, granular dataset of individual-level exercise data from more than 10 million users worldwide for about seven years to (a) measure the regularity of exercise behavior, (b) identify factors that predict a behavior continuing, (c) compare social influence in running for individuals with and without running habits, and (d) estimate the consequences of common disruptions to circumstances cues for habitual behaviors. I use modern causal inference techniques to address central questions in the psychology of habits with applications to interventions — especially social interventions — to influence exercise behavior and adoption of consumer exercise products. 2 - Location-based Advertising And Contextual Mobile Targeting Dominik Molitor, New York University, dmolitor@stern.nyu.edu Understanding how location-based advertising (LBA) can be utilized to increase sales in stores is important for offline retailers. LBA is a means to target users by making use of their location via GPS-enabled smartphones. Further, the ubiquitous nature of smartphones increases the importance of additional contextual factors such as time and weather. In particular, we analyze how contextual factors can be used to improve the prediction of responses to mobile promotions by applying unique GPS data. In particular, we examine the interplay between location, time, weather as well as co-location and users’ responses to mobile promotions. 3 - The Effect Of Referral Source On News Article Readership And Sharing Patterns Sagit Bar-Gill, Massachusetts Institute of Technology, Cambridge, MA, United States, sbargill@mit.edu, Shachar Reichman, Xitong Li The ongoing transition to online and mobile news consumption is both a challenge and an opportunity to news providers. Readers are consuming more content online, and are increasingly relying on third-party aggregators and social media to find what content to read. We employ analytic tools and fine grained news consumption data to study the effect of online referral sources on readership and sharing patterns on the Christian Science Monitor website. We explore differences in traffic patterns coming from social media and news aggregators, and examine whether the effects of referral source differ for mainstream compared to niche content. 4 - Trade-offs In Digital Advertising: Modeling And Measuring Advertising Effectiveness And Annoyance Dynamics Vilma Todri, New York University, vtodri@stern.nyu.edu Anindya Ghose, Param Vir Singh This study captures the trade-offs between effective and annoying digital advertising exposures. A hidden Markov model (HMM) is proposed that allows us to investigate the extent to which display advertising has an enduring impact on consumers’ purchase decision and whether display advertising can stimulate annoyance to consumers; we provide a conceptual framework for understanding whether persistent digital display advertising exposures constitute a mechanism of annoyance. We also study the structural dynamics of the effective and annoying display advertising effects by allowing the corresponding effects to be contingent on the latent state of the funnel path consumers reside. 211-MCC Text Analysis within Social Media Invited: Social Media Analytics Invited Session Chair: Fay Cobb Payton, North Carolina State University, Campus Box 7229, Raleigh, NC, 27695, United States, fay_payton@ncsu.edu 1 - Text Analytics - The Power Of Storytelling FayCobb Payton, Professor, Information Systems, NCSU, fay_payton@ncsu.edu Numbers do not lie. This is a typical framework for positivists (often quantitative) researchable questions. This session will provide the introduction and a case study of why text analytics can be a powerful tool for complex, often unstructured data sources. A following session will provide insights into incorporating text analytics into organizational and research objectives. SD49

110