Informs 2015

215 / 552

INFORMS Philadelphia – 2015
213
2 - How Our Networks Shape Our Privacy
Yotam Shmargad, University of Arizona, 1515 E. First St., Tucson,
United States of America,
In this study, I relate characteristics of people’s networks to the level of privacy
they experience in their social environments. I analyze over half a million users
with nearly 40 million connections on a social network site, and show that
characteristics of users’ networks can be used to predict various behaviors on the
site – including decisions to share and consume information. In particular, users
with networks containing several distinct social groups are more active on the
site.
3 - Inter-firm Managerial Social Ties, IT Supplier Selection and IT
Standardization
Oliver Yao, George N. Beckwith, Professor, Lehigh University, 621
Taylor Street, Bethlehem, PA, 18015, United States of America,
, Ling Xue, Ke Yang
We empirically test links between inter-firm managerial social ties (IMST) and IT
supplier selection and IT standardization. We find that: (1) A firm is more likely to
use an IT vendor if the firm has more IMST with the IT vendor. (2) A firm with
more IMST with its potential IT vendors uses more IT vendors. (3) More IT
vendors is associated with lower IT standardization for the firm, and such
relationship is strengthened when the firm has a greater number of IMST with its
IT vendors.
4 - Latent Space Inference of Internet-Scale Networks
Junming Yin, University of Arizona, Department of MIS,
Tucson, AZ, 85721, United States of America,
, Qirong Ho, Eric Xing
The rise of internet-scale networks with hundreds of millions to billions of nodes,
presents new scientific opportunities, such as overlapping community detection to
discover the structure of the internet. However, many existing models are difficult
or impossible to deploy at these massive scales. We propose a scalable approach
for overlapping community detection in internet-scale networks, and we
demonstrate our method on real networks with up to 100 million nodes and
1000 communities.
MC25
25-Room 402, Marriott
Data-Driven Research on Economics of Digitization
Sponsor: Information Systems
Sponsored Session
Chair: Hossein Ghasemkhani, Assistant Professor,
Purdue University, 425 W. State Street, West Lafayette, IN, 47907,
United States of America,
1 - Predicting Buying Opportunity in Retail Market with
Machine Learning
Warut Khern-Am-Nuai, Purdue University, 403 W. State Street,
West Lafayette, IN, 47907, United States of America,
, Karthik Kannan, Hossein Ghasemkhani
Previous literature has shown that many machine learning techniques are
effective in predicting stock price. However, it is not clear if those practices can be
applied to a non-financial context or not. This paper employs three machine
learning algorithms: ANNs, SVMs, and MARS to predict buying opportunities of
products in a retail market. The preliminary results suggest that machine learning
could be one potential avenue to help managers in optimizing buying decisions.
2 - Dynamic Estimation of Peer Effects and Product Engagement
Daniel Rock, Doctoral Candidate, MIT Sloan School of
Management, 30 Memorial Drive, Office 341, Cambridge, MA,
02142, United States of America, Sinan Aral,
Sean Taylor
After product adoption, consumers make decisions about continued use. These
choices can be influenced by peer decisions in networks, but identifying causal
peer influence effects is challenging. Using engagement data for Yahoo Go, a
mobile application, we apply a dynamic version of the Bramoullé et al. (Journal
of Econometrics 2009) identification strategy to estimate usage peer effects. We
compare the performance of a variety of prediction models for the instrumental
variables “first stage”.
3 - Information Technology and the Rise of the Power Law Economy
Guillaume Saint-jacques, PhD Candidate, MIT Sloan School of
Management, 100 Main St, E62-459, Cambridge, MA, 02142,
United States of America, , Erik Brynjolfsson
We show that the dramatically increasing share of income going to top earners
can be explained by the rise of the “power law economy” and argue this reflects
increased digitization and networks. Specifically, tax data (1960-2008) show that
more individual incomes are drawn from a power law, as opposed to the long-
established log-normal distribution. We present a simple theoretical model to
argue that the increased role of power laws is consistent with the growth of
information technology.
4 - The Value of Live Chat in Online Purchase
Xue Tan, University of Washington, Seattle, WA, United States of
America, Youwei Wang, Yong Tan
In today’s competitive online marketplace, adopting a live chat tool is widely
considered by merchants as a way to conduct one-to-one selling like in physical
store. By allowing customer representatives to talk to potential buyers, e-tailer
can answer consumers’ questions and decrease the level of information
asymmetry. This paper empirically examine the role of live chat in terms of
purchase conversion.
MC26
26-Room 403, Marriott
Academic Job Search Panel
Cluster: INFORMS Career Center
Invited Session
Chair: Beril Toktay, Georgia Tech, Atlanta, GA,
United States of America,
1 - Academic Job Search Panel
Moderator: Beril Toktay, ,
Panelists: Kris Johnson Ferreira, H. Edwin Romeijn,
Wedad Elmaghraby, Gad Allon
The panel will discuss the academic interview process and do’s and don’ts
associated with the job search. In addition to comments by current and former
search chairs, time will be provided for questions and answers.
MC27
27-Room 404, Marriott
Multi-objective Design Problems
Sponsor: Multiple Criteria Decision Making
Sponsored Session
Chair: Diclehan Tezcaner Ozturk, Dr., TED University, Industrial
Engineering, Ankara, Turkey,
1 - A Control Chart Recommendation System
Sidika Tunc, Research Assistant, Middle East Technical University,
Cankaya, Ankara, Turkey, , Gulser Koksal
An approach is developed to recommend the most appropriate control chart to a
novice decision maker in statistical process control. The chart selection problem is
formulated as an MCDM problem. Overall desirability of each chart is
determined. Expert knowledge is utilized. The system is tested and calibrated by
statistical experiments.
2 - Estimating Non-additive Value Functions with Active Learning in
the Ordinal Classification Setting
Levent Eriskin, Middle East Technical University, Industrial
Engineering Department, Ankara, Turkey,
, Gulser Koksal
Preference modeling is used to represent Decision Maker’s subjective preference
structure. Preference structure having some kind of interaction among criteria is
hard to model. In this study, we present results of analyses conducted for
estimating non-additive value functions having interaction structure by utilizing
active learning techniques in the ordinal classification setting.
3 - Interactive Mean-variance-covariance Optimization for
Two Responses
Melis Ozates, Research Assistant, Middle East Technical
University, Universiteler Neighborhood, Dumlupinar Avenue
No.1, Ankara, 06800, Turkey, ,
Gulser Koksal, Murat Koksalan
We develop an interactive approach for the two-response product and process
design optimization problem, explicitly considering decision maker preferences
and allowing for correlated responses. We use several performance measures to
represent the objectives that facilitate effective communication with the decision
maker.
MC27