Shu-Hsing Li

Social Media Analytics and Intelligence

In a broad sense, social media refers to a conversational, distributed mode of content generation, dissemination, and communication among communities. Different from broadcast-based traditional and industrial media, social media has torn down the boundaries between authorship and readership, while the information consumption and dissemination process is becoming intrinsically intertwined with the process of generating and sharing information. This special issue samples the state of the art in social media analytics and intelligence research that has direct relevance to the AI subfield from either an methodological or domain perspective.

Credit Rating Change Modeling Using News and Financial Ratios

Credit ratings convey credit risk information to participants in financial markets, including investors, issuers, intermediaries, and regulators. Accurate credit rating information plays a crucial role in supporting sound financial decision-making processes. Most previous studies on credit rating modeling are based on accounting and market information. Text data are largely ignored despite the potential benefit of conveying timely information regarding a firm’s outlook. To leverage the additional information in news full-text for credit rating prediction, we designed and implemented a news full-text analysis system that provides firm-level coverage, topic, and sentiment variables. The novel topic-specific sentiment variables contain a large fraction of missing values because of uneven news coverage. The missing value problem creates a new challenge for credit rating prediction approaches. We address this issue by developing a missing-tolerant multinomial probit (MT-MNP) model, which imputes missing values based on the Bayesian theoretical framework. Our experiments using seven and a half years of real-world credit ratings and news full-text data show that (1) the overall news coverage can explain future credit rating changes while the aggregated news sentiment cannot; (2) topic-specific news coverage and sentiment have statistically significant impact on future credit rating changes; (3) topic-specific negative sentiment has a more salient impact on future credit rating changes compared to topic-specific positive sentiment; (4) MT-MNP performs better in predicting future credit rating changes compared to support vector machines (SVM). The performance gap as measured by macroaveraging F-measure is small but consistent.

Effects of Differential Tax Rates on Transfer Pricing

The purpose of this paper is to study transfer pricing under asymmetric information and taxation. In accordance with the empirical evidence documented in accounting literature, this paper assumes that the firm uses one pricing system instead of two pricing systems—one for the tax purposes and the other for internal control. We provide a closed-form solution for the optimal mechanism under a dual-price system, which allows for the price credited to the manufacturing division to not equal the price charged to the distribution division. The equilibrium outcomes of the analysis suggest several interesting findings. Under a dual-price system, both divisional accounting profits at equilibrium change in the same direction with respect to the change of tax rate. However, the direct effect is larger than the indirect effect. Under a dual-price system, the division with the lower tax rate should be credited more profits than the division with the higher tax rate, but it would not fully bear all the profits.

Enterprise risk and security management: Data, text and Web mining

The Internet, Web 2.0, consumer-generated social media, and new advanced data, text, and Web mining techniques have created tremendous business opportunities. However, the potential risk and security concerns are also equally alarming. Research of relevance to enterprise risk and security assessment and analysis has gained significant interests among MIS, CS, and business researchers. Through large-scale Web enabled content collection (e.g., corporate reports, news, consumer feedback, corporate blogs and forums, and brand sentiment) and advanced mining techniques, companies and industries will be better positioned to identify potential risks and security concerns. This special issue aimed at archiving a collection of research papers of practical and novel applications, techniques, algorithms, methods, and practices in data, text, and Web mining that will make a contribution to knowledge in enterprise risk and security management. We received 33 submissions and each paper was reviewed by two to three experts in the area. After two rounds of review, 11 highquality papers were accepted. These papers report on latest research relevant to enterprise risk and security management. The first four papers of the special issue investigated the effects of news coverage and announcements, such as those on information security investment and phishing attacks, on a companys stock price and volatility. Chai, Kim, and Rao found substantial support for their hypotheses based on the public announcements of information security investment over a 10 year period from 1997 to 2006. Despite the fact that investments in data and information security are unavoidable expenses for firms, it is difficult to measure the direct return from IT security investments. This study selected the event methodology to investigate whether information security investment announcements would affect the stock price in the market. Due to the fact that phishing attack causes financial loss and shatters the confidence of customers in conducting e-commerce, Chen, Bose, Leung and Guo adopted a hybrid approach that used text phrase extraction and supervised classification to predict the severity of a phishing attack according to its risk level or financial loss generating potential. Results indicated that the classification accuracy of the hybrid approach was quite superior, and demonstrated that the key identifying variables for risk level and potential financial loss of phishing attacks were different from each other. The usefulness of accounting numbers has been an important issue for accounting researchers and general investors. However, other information sources such as financial news may also contain useful information. Chen and his colleagues investigate how financial news impacts the return-earnings relation. The news articles in the Wall Street Journal from August 1999 through February 2007 were used to construct measures for news coverage on S&P 500 companies. This study highlighted the importance of financial news in conveying value-related information to the markets.

Time-to-Event Predictive Modeling for Chronic Conditions Using Electronic Health Records

Although electronic health records (EHRs) hold promise for supporting clinical decision making, few studies have used them to model the progression of chronic conditions. To examine the feasibility of EHR-based predictive models for chronic conditions and to mitigate the associated data challenges, the authors develop a time-to-event predictive modeling framework consisting of five analytical steps: guideline-based feature selection, temporal regularization, data abstraction, multiple imputation, and extended Cox models. Using concept- and temporal-abstracted features, the proposed model attained significantly improved performance over the model using only base features.

Healthcare predictive analytics for risk profiling in chronic care: a Bayesian multitask learning approach

Clinical intelligence about a patients risk of future adverse health events can support clinical decision making in personalized and preventive care. Healthcare predictive analytics using electronic health records offers a promising direction to address the challenging tasks of risk profiling. Patients with chronic diseases often face risks of not just one, but an array of adverse health events. However, existing risk models typically focus on one specific event and do not predict multiple outcomes. To attain enhanced risk profiling, we adopt the design science paradigm and propose a principled approach called Bayesian multitask learning (BMTL). Considering the model development for an event as a single task, our BMTL approach is to coordinate a set of baseline models—one for each event—and communicate training information across the models. The BMTL approach allows healthcare providers to achieve multifaceted risk profiling and model an arbitrary number of events simultaneously. Our experimental evaluations demonstrate that the BMTL approach attains an improved predictive performance when compared with the alternatives that model multiple events separately. We also find that, in most cases, the BMTL approach significantly outperforms existing multitask learning techniques. More importantly, our analysis shows that the BMTL approach can create significant potential impacts on clinical practice in reducing the failures and delays in preventive interventions. We discuss several implications of this study for health IT, big data and predictive analytics, and design science research.

Optimal transfer pricing schemes for work averse division managers with private information

Abstract This paper models a decentralized firm under information asymmetry and effort disutility on the part of managers. We assume that managers choose efforts before observing some private information. However, after the effort choice managers receive private information on their cost parameters which they report to the headquarters. There exist many situations in which managers need to take efforts before obtaining private information; for example, the regular maintenance effort on the machine, the effort on RD 2) under certain demand conditions managers cannot receive any information rent benefit for their private information even if they have the option to renege on the contract after obtaining their private information.

The Effect of Foreign Ownership Restrictions on the Price Dynamics of Depositary Receipts—Evidence from the Taiwan and Hong Kong Markets:

In this paper, we study depositary receipt prices in regulated markets and free-entry markets. Because of their unique environments, the Taiwanese and Hong Kong markets provide interesting settings that have not yet been explored in the literature. In particular, we focus on the following: (1) the difference in the long-term price relationships between depositary receipts and underlying securities in free-entry and regulated areas, (2) the price dynamics of the depositary receipts for firms with and without the long-term equilibrium relationships between depositary receipts and underlying securities, and (3) the incremental information content of the qualified foreign institutional investor (QFII) ownership ratio for the depositary receipts issued by Taiwanese firms. The empirical results reveal that long-term equilibrium relationships between depositary receipts and underlying security prices exist for firms listed in Hong Kong, a free-entry market, but do not necessarily exist for firms listed in Taiwan with foreign ownership restrictions. The long-term equilibrium relationships between depositary receipts and underlying securities and the local market conditions are the most important factors in explaining the depositary receipt returns when the equilibrium exists. In the absence of equilibrium, the lagged returns of depositary receipts or underlying securities and the local market conditions become important. In addition, QFII ownership ratios significantly explain the depositary receipt price variations in a regulated market, which implies the restriction effect on the price dynamics of the depositary receipts.

DiabeticLink: An integrated and intelligent cyber-enabled health social platform for diabetic patients

Given the demand of patient-centered care and limited healthcare resources, we believe that the community of diabetic patients is in need of an integrated cyber-enabled patient empowerment and decision support tool to promote diabetes prevention and self-management. Most existing tools are scattered and focused on solving a specific problem from a single angle. DiabeticLink offers an integrated and intelligent web-based platform that enables patient social connectivity and self-management, and offers behavior change aids using advanced health analytics techniques. DiabeticLink released a beta version in Taiwan in July 2013. The next versions of the DiabeticLink system are under active development and will be launched in the U.S., Denmark, and China in 2014. We describe the system functionalities and discuss the user testing and lessons learned from real-world experience. We also describe plans for future development.

Product market competition and firms’ narrative disclosures: evidence from risk factor disclosures

This study examines how product market competition affects firms’ narrative disclosures of Item 1A Risk Factors in 10-K filings. We find that firms in more concentrated industries tend to disclose a greater quantity of narrative risk information. Besides, such firms provide risk disclosures more similar to those of their competitors, hence reducing the quality of the disclosure. We also document similar findings for idiosyncratic risk disclosure, which is inherently more firm-specific. The results imply that firms in more concentrated industries avoid divulging risk information in their narrative disclosures by disclosing more similar information rather than by reducing the amount of risk disclosure.