2016 INFORMS Annual Meeting Program

MC51

INFORMS Nashville – 2016

MC48

In this study, we propose an econometric model to examine the drivers of social media engagement. Set in the context of national elections, we examine the impacts of tweeting behavior of the candidates contesting in an election on the social media engagement with their constituents. To meet our objectives, we assemble a novel candidate-level data of social media engagement. 2 - The Effect Of “online following” On Contributions To Open Source Communities Xiaowei Mei, University of Florida, xiaowei.mei@warrington.ufl.edu, Mahdi Moqri, Liangfei Qiu, Subhajyoti Bandyopadhyay We estimate the effect of “online following,” a basic form of online social interaction, on members’ contributions in open source software (OSS) communities. We employ a panel vector autoregression model using individual level data across time in GitHub to achieve identification of causal effects, while controlling for individual heterogeneity and time effects. We find that the following behavior of others has a significant positive effect on users’ contribution level to the platform even after controlling for the aforementioned factors. On the other hand, users’ contribution level also has a significant positive effect on their following behavior. 3 - Manipulation For Competition: An Empirical Investigation Of Click Farming Jingchuan Pu, University of Florida, jingchuan@ufl.edu Liangfei Qiu, Hsing K Cheng Anytime there’s a monetary value added to clicks, there’s going to be people going to the dark side. This research focuses on the economic incentive of the click farming, a popular form of click fraud which is widely conducted by sellers or content generators. We cooperate with a website that uses an algorithm to detect the robot or unreal viewing activity, a black list is built for the suspicious user account which may be used as click farming. Using the detection results as a proxy of the content generators’ click farming activity, we test the impact of potential incentive, like the status of this content, content generator and the competition environment on the content generators’ click farming activity. 213-MCC Pricing and Revenue Management Applications Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: N. Bora Keskin, Duke University, Durham, NC, United States, bora.keskin@duke.edu 1 - Pricing Reservations: Dealing With No-shows Jaelynn Oh, University of Utah, jaelynn.oh@business.utah.edu We study two remedies to deal with reservation no-shows: charging no-show fees and price discrimination. We also study how to allocate capacity between reservations and walk-ins. 2 - Trade Credit And Lifetime Value Of A Newsvendor Buyer Meisam Soltani-koopa, Queen’s University, Kingston, ON, Canada, 15msk3@queensu.ca, Yuri Levin, Mikhail Nediak, Anton Ovchinnikov Trade credit typically appears as a grace period for invoice payment. It helps retailers overcome temporary cash shortages as an alternative to seeking financing from banks. We consider a supply chain with one supplier and one repeated newsvendor retailer where the supplier maximizes its long-term profit by offering the retailer a financing and wholesale price contract. In each period, the retailer decides about the quantity to order and the amount of money to borrow from the supplier or a bank. We study the lifetime value of the retailer using a dynamic Stackelberg game, from the perspective of the supplier who, as a leader, takes into account the long-term view of its relationship with the retailer. 3 - Asymptotically Optimal Markdown Policies For Demand Learning Hongfan Chen, University of Chicago, hongfan.chen@chicagobooth.edu, N. Bora Keskin Consider a firm selling a product to a population of customers with heterogeneous valuations. In this setting, we develop guidelines for designing markdown policies and derive asymptotically optimal performance guarantees. (Joint work with Bora Keskin) MC51

210-MCC New Insights from Social Media: Empirical Study, Field Experiment, and Algorithm Development Invited: Social Media Analytics Invited Session Chair: Yen-Yao Wang, Michigan State University, 632 Bogue Street, Room N204, Okemos, MI, 48824, United States, wangyen@broad.msu.edu 1 - Social Media Engagement, Product Evaluations, And Demand Spillover In The Automobile Industry Yen-Yao Wang, Michigan State University, wangyen@broad.msu.edu, Anjana Susarla, Roger Calantone, Yingda Lu Online Word of Mouth (WOM) is an important aspect of consumer-firm relationship and a leading indicator of product performance. However, prior research focuses considerably on the static view of it. This paper examines the dynamics of the spillover effects in online WOM in the US automobile industry. To measure online WOM, we focus specifically on customers’ test drive experience. We collected data from around 1000 different social media platforms for 32 car brands from 2009 to 2015. We used a Bayesian modeling framework and estimated the model using Markov Chain Monte Carlo (MCMC) methods. 2 - The Dark Side Of Positive Social Influence Che-Wei Liu, University of Maryland, College Park, MD, Un ited States, cwliu@rhsmith.umd.edu, Ritu Agarwal, Guodong (Gordon) Gao Social influence has been widely used to transform behaviors. However, when individuals are pressured to conform to behavior of the majority, would it lead to an unexpected backfire effect? We conducted a randomized field experiment of more than 10,000 individuals for a two-month period on an online physical activity community to examine the dark side of social influence. We studied the effect of social norms on users’ goal setting and goal achievement behavior. While social influence increased the rate of goal setting, strikingly, we also observe the dark side of social influence. Our findings have important implications for the design of interventions in the context of mHealth technologies. 3 - Influencers In Social Media – An Assessment Of Algorithmic Approaches In Big Data Environments Shih-Hui Hsiao, Lawrence Technological University, Southfield, MI, United States, suade0904@msn.com, Ram Pakath Growing social media usage, accompanied by explosive growth in related Big Data, has resulted in increasing interest in finding automated ways of discovering influencers (i.e., opinion leaders) in online social interactions. Beginning 2008, about two dozen variants of six basic approaches have been proposed. Yet, there is no comprehensive study investigating the relative efficacy of these methods in specific settings. We investigate and report on the relative performances of 15 methods on 9 twitter data sets ranging in size from tens of thousands to hundreds of thousands of tweets. 4 - Natural Language Processing: From Text Mining To Social Media Analysis Chih-Hao Ku, Assistant Professor, Lawrence Technological University, 21000 W 10 Mile Rd, M309A, Southfield, MI, 48331, United States, cku@ltu.edu, Yung-Chun Chang Today, the number of digital reports, e.g., crime reports and social media data derived from Twitter, LinkedIn, and Facebook are growing rapidly. However, this immerse amount of digital data post challenges and opportunities for data analysis. The rise of social media has drawn interest on text mining and social media analysis, e.g., sentiment analysis. Natural language processing (NLP) is an important component for those analyses. We report here on research work on text mining, sentiment analysis, and future trend using NLP techniques.

MC49

211-MCC Estimating Sentiments Using Social Media Invited: Social Media Analytics Invited Session

Chair: Subodha Kumar, Mays Business School, Texas A&M University, 301-F, Wehner, 4217 TAMU, Mays Business School, College Station, TX, 77843, United States, skumar@mays.tamu.edu 1 - The Effects Of Social Media Sentiment On Engagement Rakesh Reddy Mallipeddi, Mays Business School, Texas A&M University, College Station, TX, United States, rmallipeddi@mays.tamu.edu, Ramkumar Janakiraman, Subodha Kumar, Seema Gupta

199