Research

The Pamplin College’s Center for Business Intelligence and Analytics (CBIA) serves as an interdisciplinary resource to support faculty research both within the College and in cooperation with other Centers and academic units on campus, curricular initiatives for students with interests in business intelligence and analytics, and outreach to the Virginia business community. CBIA research focuses on the application of systematic analysis (both quantitative and qualitative) to vast collections of business data in order to leverage it for business planning and decision making has assumed a central place in management

Business intelligence and analytics (BI&A) refers to the skills, technologies, applications, processes, and practices which businesses use to gain insights and make informed and optimal decisions in all areas.  BI&A has been used by businesses around the world to cut costs, gain insights about products, customers, business operations, and operational environment. It has shown tremendous value for businesses’ performance excellence.

Business analytics makes extensive use of data of all sizes, quantitative methods (statistics, operations research/management science, machine learning/deep learning, and artificial intelligence), explanatory and predictive models, and data-driven management to drive business decision-making.

Pamplin college of Business faculty members have accumulated extensive expertise in all BI&A areas, especially in text analytics, quality analytics, social media analytics, predictive analytics, supply chain analytics, consumer intelligence, competitive intelligence, deep learning, image and video analytics, descriptive and prescriptive analytics. Our faculty members have applied BI&A techniques for supply chain risk management, product and service issue identification, online review mining, crisis management, consumer profiling, market intelligence, healthcare operation optimization, capital markets, pharmacovigilance, operation management, online community mining, smart travel and tourism, etc.

Faculty Research

  • “The Proximity Paradox: Why People Donate Less to Those “Close to Home”, J Jiang

A review of how geographic distance affects donations using a data collected from DonorsChoose.org. Findings: a counterintuitive result that people donate less to beneficiaries in close proximity (e.g., in the same zip code) than more distant ones (e.g., those in different zip codes).

  • “The Role of Feedback in Crowdsourcing Contest: A Theoretical and Empirical Study”, J Jiang

Development of a game-theoretic model of feedback in innovation contests. We show that feedback plays an informational role in mitigating the information asymmetry between the seeker and solvers, thereby inducing solvers to increase their effort level. We also show that the incentive mix of the contest (including contest reward, solver expertise, and solver population) has a direct effect on solver effort; that is, solvers’ effort level increases with the contest reward and solver expertise but decreases with solver population.

Interestingly, by endogenizing the seeker’s choice of giving feedback, we find that the seeker gives out more feedback comments when contest reward is higher, solver expertise is higher and solver population is smaller. Thus, the incentive mix also has an indirect effect on solvers’ effort level through feedback volume, indicating that feedback also has a mediating role in the relationship between the incentive mix and solvers’ effort.

  • “Do Customer Reviews Matter in Product Innovation – An Empirical Study of the Google Play Store”, Z. Qiao, A. Wang, P. Fan

This study investigates how customer knowledge impacts product innovation using data collected from Google Play store. Guided by Elaboration Likelihood Model of persuasion theory, our research shows that long and easy-to-read user reviews with mildly reviews can increase the frequency of product innovation activities. Our findings highlight the need for researchers to explore user reviews in the context of customer-driven product innovation. (Runner for Best Paper Award in Pre-ICIS SIGDSA Workshop)

  • “Identify Product Defects from User Complaints Using a Bayesian Mixture Model”, X. Zhang, Z. Qiao, P. Fan, A. Wang

This research develops a Bayesian Mixture Model to identify the most complained issues of a product from vast volume of customer reviews. Product defect is taken as a latent variable, and user reviews are used as observations in this model. Multinomial distribution and Gaussian distribution are incorporated to formulate the key entities of customer reviews, such as symptom words and incident date. EM algorithm is applied to infer the probability of defects, their most probable associated entities, and the representative complaints of defects. The evaluation results demonstrate that the proposed model is effective in finding essential product defect information from a large number of user complaints.

  • “Project Communications on Crowdfunding Success: An Empirical Study Based on Elaboration Likelihood Model”, J. Zhou, P. Fan, A. Wang, L. Wallace

This study introduces a dual-processing theory – the Elaboration Likelihood Model (ELM) – into crowdfunding domain. We employ ELM and text mining techniques to investigate influential characteristics related to funding success and to better understand the information processing procedures used by backers to make funding decisions. First, we identify three argument quality variables (length, readability, and sentiment) and two source credibility variables (past experience and past expertise) that are influential on funding success through central and peripheral routes. Second, we find a positive interaction effect on funding success between argument quality and source credibility. Third, we find a substitution effect on funding success between information disclosed in text and multimedia content. Fourth, we find that backer involvement has positive moderating effects on the both central and peripheral routes. Our results are robust across different measures of funding success (funding status, funding ratio, and funds pledged) and a variety of settings. Additionally, we reconcile the mixed results in the literature regarding the influence of duration on funding success. Specifically, we find that duration has positive impacts on successful projects but negative impacts on failed projects, suggesting that successfully funded projects have different behavior than failed projects.

  • “An Examination of the Readability of Tax Footnotes: Determinants and Implications for the Valuation of Tax Avoidance”, P. Fan, K. Inger, M. Meckfessel, J. Zhou

This paper investigates the information provided to stakeholders in corporate disclosures by examining the readability of tax footnotes. We find a positive association between tax avoidance and tax footnote readability for firms with relatively low levels of tax avoidance, suggesting managers highlight good performance by providing straightforward disclosures and conceal poor performance by providing information that is difficult to process. In contrast, there is no association between tax avoidance and tax footnote readability in firms with relatively high levels of tax avoidance. Consistent with these results, we find that investors discount (place a premium on) tax avoidance when the tax footnote is harder to read in low (high) avoidance firms. We provide evidence that the readability of the tax footnote has informational value beyond the overall readability of the annual report.

  • A Comparative Analysis of Major Online Review Sites: Implications for Social Media Analytics in Hospitality and Tourism”, P. Xiang, Y. Ma, Q. Du, P. Fan

The growing impact of user-generated content especially online reviews has been widely documented in hospitality and tourism. However, existing studies tend to use single data sources and the quality of data is anecdotal or based upon popularity of the websites. This project compares information quality of three major review sites, namely Tripadvisor, Expedia and Yelp, using the entire hotel population in Manhattan, NYC. We examined information quality in terms of the site’s representation of the hotel product from both the supply and demand sides. Online reviews’ linguistic features, prominent topics, distribution of helpful reviews, review features that contribute to satisfaction rating are compared across the three sites. Results clearly show the consistency and discrepancies in consumers’ experience and evaluation of the hotel product. This study offers implications for developing social media analytics in hospitality and tourism.

I. Lanham, M., Badinelli, R. “Developing a Rebalancing Parameter Table for Binary Classification Modeling.”

2. Lanham, M., Badinelli, R. “Merging Business KPIs With Predictive Model KPTs for Binary Classification Model Selection.”

I. Abrahams, A., and C. Ragsdale, “A Decision Support System for Patient Scheduling in Travel Vaccine Administration,” Decision Support Systems, Vol. 54, No. 1, pp. 215-225, 2012.

I. Winkler M, Abrahams AS, Gruss R, and Ehsani J (2016). Toy Safety Surveillance from Online Reviews. Decision Support Systems, Forthcoming. Accepted 20 May 2016.  5-year ISI Impact Factor (2015): 3.271

2. “Computational Rhetoric via Sequence Alignment: An Innovative Text Analytics Approach for Quantifying Persuasion” Şeref, MMH, Şeref, O, Abrahams, A, Warnick, Q, Hill, S. (Presented at the 2014 CBIA Workshop)

We develop a new methodology to identify rhetorical moves in large text data. We perform an initial qualitative analysis of a random sample of text to train our tagging algorithm. We are then able to quickly tag large amounts of text and therefore quantify the number of moves for different types of persuasion. This quantitative data can then be used for classification approaches in understanding and predicting decision behaviors. We demonstrate our method by analyzing pitch arguments used to justify stock decisions in an online community.

3. “Misdirection in Tax Reports for Corporate Inversion” MMH Seref, D Salbador. (Presented at the 2015 CBIA Symposium on Analytics)

We use new developments in text analytics, including computational rhetoric, to identify patterns of language in tax reports that may indicate compliance or fraud.

4. “Computational Rhetoric via Sequence Alignment: An Innovative Text Analytics Approach for Quantifying Persuasion” Şeref, MMH, Şeref, O, Abrahams, A, Warnick, Q, Hill, S.

5. “Mapping the Defect Genome: Improving Automated Defect Detection from Big Data using Techniques from Computational Biology” A Abrahams, O Seref, MMH Seref.

6. “A Song is Worth 1000 Words: Evaluating Potential Country Song Hits using Computational Rhetoric” C Ragsdale, MMH Seref, O Seref, A Abrahams.

7. “Critical Analysis of Accounting Reports: Do the Words Match the Numbers?” MMH Seref, J Brozovsky.

8. “Impact of Financial Media on IPO Investment Decisions” MMH Seref, R Barkhi.

9. “Misdirection in Tax Reports for Corporate Inversion” MMH Seref, D Salbador.