2016 INFORMS Annual Meeting Program

MC06

INFORMS Nashville – 2016

MC06 102A-MCC Text Analytics for Quality Management Sponsored: Data Mining Sponsored Session Chair: Alan Abrahams, Virginia Tech, Pamplin 1007, Blacksburg, VA, 24061, United States, abra@vt.edu 1 - Human Intelligence In Keyword Query Formulation: Comparing The Recall Performance Of Computer- And Human-generated Smoke Word Lists Richard Gruss, Pamplin College of Business, Virginia Tech, Blacksburg, VA, United States, rgruss@vt.edu Alan Abrahams, Siriporn Srisawas Are humans better at producing safety concern keyword searches than computers? 72 subjects annotated 263 reviews of over-the-counter medicines, half of which contained known safety concerns. Subjects then compiled a list of words and phrases that might be effective filter terms in finding reviews with safety concerns. Subjects then collaborated in groups of 2, 3 or 4 to assemble a group list. Lists were scored on how many of the known safety concerns they were able to recall in a separate test set. Preliminary results indicate that human- generated lists frequently outperform computer-generated lists. 2 - Let’S Not Get Too Sentimental: A Critical Analysis Of Sentiment Analysis For Quality Surveillance Nohel Zaman, Virginia Tech, Blacksburg, VA, 24060, United States, znohel@vt.edu, Alan Abrahams, Richard James Gruss, Siriporn Srisawas Our study will be beneficial for quality management (QM) professionals analyzing unstructured user-generated-content in social media. The goal is to determine whether and to what extent negative sentiment and defect existence are associated, in different products across multiple industries. With product defects being expensive, this paper could help manufacturers more rapidly discover defects. This paper aims to assess which sentiment and non-sentiment scoring methods are most effective at finding product defects in each industry, and which methods generalize well across industries. 3 - Online Reviews To Revenue: Contributory Factors And In this study, online reviews of airlines are examined with respect to company revenue. Though previous studies have examined online reviews’ effect on product sales, the relationship has not been elicited in service industries, including the airline industry. Using passengers’ reviews of the 7 airline groups with the highest revenue we consider the impact of these reviews on the quarterly revenue of each of these companies. Compared with other major industries, the airline industry has the highest defect rate in their online reviews. Our findings suggest that online reviews do have an impact on company revenue, however this impact is dampened by offline maneuvers. 4 - Identifying Product Defects From User Complaints: A Probabilistic Defect Model Alan Wang, Associate Professor, Virginia Tech, 2070 Pamplin Hall, Virginia Tech, Blacksburg, VA, 24060, United States, alanwang@vt.edu, Xuan Zhang, Zhilei Qiao, Weiguo Fan, Edward A Fox Discovering potential product defects from large amounts of user complaints is a challenging task. In this research, we develop a probabilistic defect model (PDM) that identifies the most critical product issues and corresponding product attributes (e.g., product model-year, defective components, symptoms, etc.), simultaneously. We conduct comprehensive evaluations to ensure the quality of discovered information. Our research has significant managerial implications for managers, manufacturers, and policy makers. MC07 102B-MCC Data Analytics in Renewable Energy Sponsored: Data Mining Sponsored Session Chair: Zijun Zhang, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon Tong, Hong Kong, zijzhang@cityu.edu.hk 1 - Exploratory Data Analytics For Temporal Lighting Energy Usage In Commercial Buildings Mingyang Li, University of South Florida, Tampa, FL, United States, mingyangli@usf.edu, Kara C Heuer, Dina Villalba-Sanchez, Zhe Song Moderating Effects In The Airline Industry Zachary Davis, Virginia Tech, zached1@vt.edu Alan S. Abrahams, Lara Z Khansa

Lighting energy usage accounts for a major proportion of building energy consumption. A better understanding of lighting energy data will facilitate building energy management. Conventional studies mainly focused on aggregate- level energy usage, ignoring the temporal usage patterns featured in stochasticity, nonlinearity and intermittency. In this study, an exploratory data analytics approach is proposed to extract, cluster and visualize temporal patterns of lighting energy usage in commercial buildings. A real data study is further provided to illustrate the proposed work and demonstrate its validity. 2 - Characterization Of Air Traffic Network Using Ads-b Data Lishuai Li, City University of Hong Kong, P6606, AC1, Tat Chee Avenue, Kowloon, Hong Kong, lishuai.li@cityu.edu.hk Pan Ren Airspace capacity has been credited as a major factor for air traffic congestion and flight delays. However, few studies provided measures of airspace capacity and efficiency for a large air traffic network. This research aims at evaluating whether airspace capacity is a significant factor in relation to recent air traffic delays in China. We developed a novel method to characterize flow patterns in the airspace and construct an air traffic network using cluster analysis on historical flight trajectories. Findings will be useful in evaluating the efficiency and robustness of an air traffic network in relation to its actual operation and management. 3 - Image-based Wind Turbine Blade Surface Crack Detection And Analysis Zijun Zhang, City University of Hong Kong, zijzhang@cityu.edu.hk Long Wang A data-driven framework for automatically detecting wind turbine (WT) blade surface cracks based on images taken by unmanned aerial vehicles (UAVs) is proposed in this paper. Haar-like features are applied to depict crack regions and train a cascading classifier. The computational results demonstrate that the proposed framework can successfully provide the number of cracks and locate them in original images. MC08 103A-MCC Innovation in Product and Service Development Invited: Business Model Innovation Invited Session Chair: Morvarid Rahmani, Georgia Institute of Technology, Atlanta, GA, United States, morvarid.rahmani@scheller.gatech.edu 1 - An Economic Analysis Of Customer Codesign Sreekumar R Bhaskaran, Southern Methodist University, sbhaskar@mail.cox.smu.edu, Amit Basu A key barrier to companies successfully engaging customers in the design of new products is customers fearing that they will be forced to pay much more for the custom products they help design. We show how a firm can motivate its customers to engage in co-design through its product line choices. The effect of market and firm characteristics on the value of engaging customers in the co- design process is also examined. In addition, we analyze the effects of (a) information asymmetry about the firm’s co-design capability, and (b) competition, on the firm’s decisions regarding co-design. 2 - Allocating Customer Control In Service Processes Ioannis Bellos, George Mason University, ibellos@gmu.edu, Stylianos Kavadias In most services customers actively participate in the service deliver process. In practice we observe services that require varying degrees of customer involvement. Motivated by this, we develop an analytical model to determine which parts of a service process should be performed by the service provider and which parts should be delegated to the customers. 3 - Sourcing Innovation: Private And Public Feedback In Contests Contests, in which contestants compete for a prize offered by a contest holder, have become a popular way to source innovation. Despite great interest from the academic community, many important managerial aspects of contests have received very little formal inquiry. The most important of these is feedback from the contest holder to the contestants while the contest unfolds. This paper sets out to establish a comprehensive understanding of how to give feedback in a contest by answering the questions of when to give feedback and when not to give feedback and which type of feedback to give, public (which all solvers can observe) or private (which only the concerned party can observe). Jurgen Mihm, Insead, Fontainebleau, France, jurgen.mihm@insead.edu, Jochen Schlapp

184