Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MB54

3 - Revenue from Matching Platforms James Schummer, Northwestern University, 2211 Campus Dr, Evanston, IL, 60208, United States, Philip Marx We consider matching platforms on which agents form pairs. The platform commits to a stable matching mechanism, charging fees to both sides. Agents on the short side of such markets capture more value than those on the long side (Ashlagi et al. 2017). Nevertheless we show that the platform does not price discriminate between the sides based on their relative sizes. We demonstrate that the cost of committing to stability vanishes in large markets. While preference correlation leads the platform to bias its prices in imbalanced markets, we show that the direction of price bias depends on the type of correlation; these effects are absent from models of two-sided markets without capacity constraints. 4 - On Finding Stable and Efficient Solutions for the Team Formation Problem Robert Day, University of Connecticut, 2100 Hillside Road U-1041, Storrs, CT, 06269-1041, United States, Hoda Atef Yekta, David Bergman We study a mathematical-programming approach to team formation, focused on the interplay between two of the most common objectives considered in the related literature: economic efficiency (i.e., the maximization of social welfare) and game-theoretic stability (e.g., finding a core solution when one exists). With a weighted objective across these two goals, the problem is modeled as a bi-level binary optimization problem, and transformed into a single-level, exponentially sized binary integer program. We then devise a branch-cut-and-price algorithms and demonstrate its efficacy through an extensive set of simulations, with favorable comparisons to other algorithms from the literature. n MB54 North Bldg 232B Opportunities and New Directions for Behavioral OM Sponsored: Behavioral Operations Management Sponsored Session Chair: Anyan Qi, The University of Texas at Dallas, Richardson, TX, 75080, United States Co-Chair: Ruth Beer, Indiana University, Kelley School of Business, Indiana University, Kelley School of Business, Bloomington, IN, 47404- 5180, United States 1 - Opportunities for Behavioral Research within Supply Chain Management Tava Olsen, University of Auckland, ISOM, Business School, University of Auckland, Auckland, 1142, New Zealand This talk examines opportunities for behavioral research within supply chain management (broadly defined). In particular, I will discuss models and empirical anomalies that could benefit from behavioral research (in terms of lab experiments, behaviorally inspired modelling, or empirical research). The talk concludes with a discussion my perception of the role for behavioral research going forward. 2 - Consumer Behavior in Pricing Applications Georgia Perakis, Massachusetts Institute of Technology, Sloan School of Management, MIT, 100 Main Street Rm E62-565, Cambridge, MA, 02142-1347, United States In recent years, there has been increasing interest in incorporating behavioral phenomena in Operations Management applications. In particular, in this talk, I will discuss some recent work on pricing applications where amending the traditional models to include consumer behavior (such as memory and anchoring effects) leads to interesting new insights. 3 - Queueing Systems with (MIS)behaving Servers and Customers Mor Armony, New York University, 44 West 4th Street #8-62, New York, NY, 10012, United States Traditional queueing theory has assumed that customer arrival times and service requirement as well as server service speed are exogenously determined. This assumption makes sense for systems with non-human customers and servers. In practice, in service systems that involve humans as customers and / or servers, behavioral effects may lead to violation of these assumptions. Specifically, both customers and servers may respond to incentives related to social comparisons, sense of ownership, and fatigue, to name a few. These behaviors can dramatically change the performance and optimal control of queueing systems. We survey recent literature and present opportunities for further research.

n MB52 North Bldg 231C Social Media and Information Warfare Emerging Topic: Social Media Analytics Emerging Topic Session Chair: Tauhid Zaman, Ph.D., MIT, Cambridge, MA, 02139, United States 1 - Detecting Influence Campaigns in Social Networks using the Ising Model Nicolas Guenon des Mesnards, MIT, Boston, MA, United States, Tauhid Zaman Recent news about Russian propaganda bots influencing elections has given the bot detection problem new importance. This problem can be modeled using the Ising model from statistical physics. This allows for joint detection of multiple bots. We show that the problem reduces to a max-flow problem, allowing for an efficient solution. We find that our bot-detection algorithm is able to find bots more accurately than existing state of the art methods. Our analysis of the bot tweets suggests that they are engaged in coordinated influence campaigns. 2 - Time Varying Opinion Dynamics with Stubborn Agents David S. Hunter, Massachusetts Institute of Technology, Cambridge, MA, 02139, United States We propose a social network opinion dynamics model where individuals grow more stubborn with time. We are able to prove several results on the convergence of the opinions. Depending upon how fast individuals become stubborn, the opinions may never converge, converge in mean, or converge in probability. We provide a precise characterization of these different convergence conditions. We also find that when the opinions converge, the equilibrium opinions are given by a linear system which has a form similar to Ohm’s Law from circuit theory. 3 - Optimizing Influence Campaigns with Stubborn Agents Tauhid Zaman, MIT, MA, United States Often there is a need to run an influence campaign in a social network. This can be done using “stubborn agents who shift the opinions of non-stubborn people. We use an Ohm’s Law equilibrium condition for the opinions as the basis for an integer program for the optimal placement of the agents. We prove that this problem is submodular, allowing for accurate greedy solutions. Using several simulated and real network topologies, we show that a small number of agents can have a strong influence on the population. n MB53 North Bldg 232A Matching Theory I Sponsored: Auction and Marketing Design Sponsored Session Chair: Thayer Morrill, North Carolina State University, Raleigh, NC, 27695, United States 1 - Obvious Manipulations Thayer Morrill, North Carolina State University, Campus Box 8110, Raleigh, NC, 27695, United States, Peter Troyan Game theory is based on the idea that if an agent has the incentive to deviate, then she will do so. However, it is clear that not all incentives to deviate are the same. How much higher of a payoff does the agent realize from deviating? Does a deviation expose the agent to more risk? Does it require a great deal of information to know whether or not a deviation will be profitable? For some mechanisms there is clear evidence that agents are adept at identifying opportunities for manipulation in practice, while for others, though opportunities for manipulation exist, they are observed much less frequently. This paper seeks a simple and tractable method for determining when a mechanism is easy to manipulate. 2 - Efficient and Pretty Fair Course Assignment with Quotas Stefan Waldherr, Technische Universitat Munchen, Boltzmannstr 3, Room 01.10.054, Munich, 85748, Germany, Martin Bichler, Alexander Hammerl, Thayer Morrill We focus on course assignment problems with minimum and maximum quotas. It is well known that even without minimum quotas there does not exist a strategyproof mechanism that always selects an efficient and fair assignment. While the extended seats TTC mechanism (ESTTC, i.e. TTC incorporating minimum quotas) satisfies strategyproofness and efficiency for minimum quotas, it is very unfair and leads to many instances of justified envy. In this talk, we present extensions to ESTTC which satisfy weak fairness axioms. Further, we leverage field data from a large-scale course assignment application where we can show a significant reduction of justified envy when applying our mechanisms.

167