Search

PurposeTo overcome the shortcomings of traditional industry classification systems such as the Standard Industrial Classification Standard Industrial Classification, North American Industry Classification System North American Industry Classification System, and Global Industry Classification Standard Global Industry Classification Standard, the authors explore industry classifications using machine learning methods as an application of interpretable artificial intelligence (AI).Design/methodology/approachThe authors propose a text-based industry classification combined with a machine learning technique by extracting distinguishable features from business descriptions in financial reports. The proposed method can reduce the dimensions of word vectors to avoid the curse of dimensionality when measuring the similarities of firms.FindingsUsing the proposed method, the sample firms form clusters of distinctive industries, thus overcoming the limitations of existing classifications. The method also clarifies industry boundaries based on lower-dimensional information. The graphical closeness between industries can reflect the industry-level relationship as well as the closeness between individual firms.Originality/valueThe authors’ work contributes to the industry classification literature by empirically investigating the effectiveness of machine learning methods. The text mining method resolves issues concerning the timeliness of traditional industry classifications by capturing new information in annual reports. In addition, the authors’ approach can solve the computing concerns of high dimensionality.

We study vector-valued almost minimizers of the energy functional $$\int_D\left(|\nabla\mathbf{u}|^2+\frac2{1+q}\left(\lambda_+(x)|\mathbf{u}^+|^{q+1}+\lambda_-(x)|\mathbf{u}^-|^{q+1}\right)\right)dx,\quad00$, we take the epiperimetric inequality approach and prove the regularity for both almost minimizers and the set of "regular" free boundary points., Comment: 39 pages. arXiv admin note: text overlap with arXiv:2112.00676

We report the one-pot synthesis of 4-amino thiochromans using simple aminocyclopropanes and thiophenols through a formal [3+3] annulation reaction. This reaction proceeds under mild conditions with good functional group tolerance. The thiochroman core was formed with complete regioselectivity, and modification of complex drug molecules containing an aminocyclopropane was also realized.

This study examines the dynamics of the digital divide between middle- and low-income groups of 44 African countries in the context of three technologies – mobile cellular, Internet, and fixed broa...

Background and purpose: Pyogenic spinal infection (PSI) has recently been on the rise due to aging and increasing degenerative spinal disease related procedures. PSI requires long-term antibiotic treatment and is followed by sustained functional disability even after successful treatment. This study aimed to analyze the clinical factors associated with medical burden and functional status of PSI. Methods: This retrospective study involved patients with non-postoperative PSI of thoraco-lumbo-sacral area in a single tertiary hospital. The length/cost of hospitalization with an antibiotic therapy and severity of back pain using the short form 36 (SF-36) were defined as the medical burden and functional status, respectively. We analyzed the clinical factors associated with medical burden and functional status. Results: We enrolled 142 patients (91 males and 51 females). The length and cost of hospitalization were 55.56 ± 27.09 (7–172) days and $14,070.17 ± 9289.39 (1611.87–48,722.35), respectively. A recurrence rate of 7.7% (11/142) and significant improvement of SF-36 at six months after completion of antibiotic treatment were noted (p < 0.05). Procedure-related (OR 2.702), C-reactive protein (CRP; OR 1.062), bacteremia (OR 4.966), additional surgical treatment (OR 6.524), recurrence (OR 12.453), and paraspinal abscess (OR 5.965) for above-average length of hospitalization were observed; female (OR 4.438), CRP (OR 1.071), bacteremia (OR 4.647), additional surgical treatment (OR 6.737), recurrence (OR 22.543), and extent of lesion (OR 1.431) for above-average cost of hospitalization; leg weakness (OR 15.966), white blood cell (WBC; OR 1.116), Charlson’s comorbidity index (CCI, OR 1.485), and identification of causative bacteria (OR 2.913) for below-average initial SF-36 were observed; leg weakness (OR 7.975) and WBC (OR 1.094) for below-average 6-month SF-36 were the statistically significant clinical factors in the multivariable logistic regression analysis (p < 0.05). Conclusion: Recurrence and leg weakness were identified as the most important clinical factors for medical burden and functional status in PSI, respectively. We think that it is necessary to actively suppress recurrence and manage neurological deficits for decreasing medical burden and achieving favorable functional outcome in the treatment of PSI.

In this paper we study vector-valued almost minimizers of the energy functional $$ \int_D\left(|\nabla\mathbf{u}|^2+2|\mathbf{u}|\right)\,dx . $$ We establish the regularity for both minimizers and the "regular" part of the free boundary. The analysis of the free boundary is based on Weiss-type monotonicity formula and the epiperimetric inequality for the energy minimizers., 25 pages

The Internet enables businesses to acquire a great deal of information, including prices in the open markets. In this study, we investigate what the value of reference price information is to a company in the market and how the company can make use of such information. Using business analytics, we were able to estimate prices of used cars for a rent-a-car company. The results show that a smart pricing information system is useful for collecting online reference price information and for estimating future prices of used cars and rental prices.

As game industry prospers, the negative side of games becomes highlighted as well as its contributions to economy growth. In spite of strong arguments for the necessity to regulations as a means to decrease addiction or overindulgence, research has produced future suggestions rather than quantifiable evidence. In this paper, we propose adopting a simulation approach in addition to quantitative approaches to better understand optimal regulatory levels since a simulation approach can visualize unexpected side effects of regulations. In this study, we suggest the application of an agent-based model (ABM) as a smart service to measure the effects of regulatory policies. We review cases applying ABM in various domains and consider the possibility of using an ABM to understand the effectiveness of web board-game regulations. We find that the ABM approach would be useful in several areas, such as the analysis of regulatory effects that reflect a variety of characteristics, the measurement of micro-regulatory effects, and the simulation of regulations.

In this dissertation, we consider almost minimizers for the thin obstacle problems in different settings: Laplacian, fractional Laplacian and equation with variable coefficients.In Chapter 1, we consider Anzellotti-type almost minimizers for the thin obstacle (or Signorini) problem with zero thin obstacle and establish their C1,β regularity on the either side of the thin manifold, the optimal growth away from the free boundary, the C1,γ regularity of the regular part of the free boundary, as well as a structural theorem for the singular set. The analysis of the free boundary is based on a successful adaptation of energy methods such as a one-parameter family of Weiss-type monotonicity formulas, Almgren-type frequency formula, and the epiperimetric and logarithmic epiperimetric inequalities for the solutions of the thin obstacle problem. This chapter is based on a joint work with Arshak Petrosyan [1]. In Chapter 2, we study almost minimizers for the thin obstacle problem with variable Hölder continuous coefficients and zero thin obstacle and establish their C1,β regularity on the either side of the thin space. Under an additional assumption of quasisymmetry, we establish the optimal growth of almost minimizers as well as the regularity of the regular set and a structural theorem on the singular set. The proofs are based on the generalization of Weiss- and Almgren-type monotonicity formulas for almost minimizers established earlier in the case of constant coefficients (Chapter 1). This chapter is based on recent joint work with Arshak Petrosyan and Mariana Smit Vega Garcia [2]. In Chapter 3, we introduce a notion of almost minimizers for certain variational problems governed by the fractional Laplacian, with the help of the Caffarelli-Silvestre extension. In particular, we study almost fractional harmonic functions and almost minimizers for the fractional obstacle problem with zero obstacle. We show that for a certain range of parameters, almost minimizers are almost Lipschitz or C1,β-regular. This is based on a work in collaboration with Arshak Petrosyan [3].

We describe different modes of dimerization of various α',γ-dioxyenone derivatives with potential applications to the synthesis of high-order securinega alkaloids. We learned that the relative stereochemical relationship between α'- and γ-hydroxyl groups of the α',γ-dihydroxyenone derivative determines the mode of dimerization. While cis-α',γ-dioxyenone 26 provided the Rauhut-Currier-type (RC-type) dimer 31 upon reaction with TBAF, trans-α',γ-dihydroxyenone 34 afforded dimeric tetrahydrofuran derivative 41 under the same reaction conditions. We also noticed that the protection of the γ-hydroxyl group drastically changes the reaction outcomes. While cis-α'-oxy-γ-OPiv-enone 49 did not show any reactivity in the presence of TBAF, trans-α'-hydroxy-γ-OPiv-enone 45 produced the RC-type dimer 46 under the same reaction conditions. Computational analysis revealed the detailed mechanism of the latter transformation.

Do larger cities still suffer from higher crime rates? The scaling relationship between the number of crimes and the population size for the maximum of 758 cities with more than 50,000 inhabitants in the United States from 1999 to 2014 was analyzed. For the total group of cities, the relationship is superlinear for both violent and property crimes. However, for the subgroups of the top 12, top 24, and top 50 largest cities, the relationship changes to sublinear for both violent and property crimes. Results from the panel data analysis are in support of these findings. Along with population size, income per capita and population density also influence the outcome of crime counts. Implications from these findings will be discussed.

We consider Anzellotti-type almost minimizers for the thin obstacle (or Signorini) problem with zero thin obstacle and establish their $$C^{1,\beta }$$ regularity on the either side of the thin manifold, the optimal growth away from the free boundary, the $$C^{1,\gamma }$$ regularity of the regular part of the free boundary, as well as a structural theorem for the singular set. The analysis of the free boundary is based on a successful adaptation of energy methods such as a one-parameter family of Weiss-type monotonicity formulas, Almgren-type frequency formula, and the epiperimetric and logarithmic epiperimetric inequalities for the solutions of the thin obstacle problem.

This paper examines the dynamics of the digital divide between two subgroups of middle-income and low-income African countries in the context of three ICT technologies. During the period of 2000 to 2015, the digital divide has been narrowing faster at the annual rate between 11.3% and 3.53% per year. Furthermore, we used the convergence analysis method to test for the speed of catch-up, and the results shows statistically valid catch-up process in both subgroups in all the three technologies and the low-income subgroups had faster speed of catch-up speed in all cases. When we combined the results from the narrowing digital divide and the speed of catch-up, we discovered that there is a positive relationship in all three technologies, as the narrowing digital divide was also accompanied by a faster catch-up process from lagging countries.

Do larger cities still suffer from a higher crime rates? The scaling relationship between the number of crime counts and the population size for the maximum of 758 cities with more than 50,000 inhabitants in the United States during 1999 to 2014 was analyzed. For the total group of cities, the relation is superlinear for both violent and property crime. However, for the subgroups of top 12, top 24, and top 50 largest cities, the relations change to sublinear for both violent and property crime. Results from multivariate panel data analysis are in support of these findings. Next to population size, income per capita and population density have also influenced the outcome of crime counts. Implications from these findings will be discussed.

Shortly after Groupon started its business in 2008, selling one deal a day with substantial price discounts, daily-deal sites became new online shopping places for many people. Starting with Groupon, most daily-deal sites required that voucher sales be higher than a predetermined number before deals become active. This feature, known as the “tipping point,” was a unique characteristic of the daily-deal business and is identified as one of the most prominent features of social shopping. Most daily-deal sites also required that a redemption period start after a deal was over and be fixed, usually 90 days, presumably to maximize the promotional effect of deals by encouraging rapid voucher redemption. The question remains, however, whether such features actually contributed to the success of the daily-deal industry. Using individual-level panel data from a major daily-deal site in Korea, we analyze whether consumers' purchase and redemption behaviors were affected by these features and how consumers changed their behaviors as they continued to purchase and redeem vouchers over time. We find that the presence of the tipping point did not boost voucher sales and likely deterred new customers from buying deals right away. We also find that new customers tended to redeem their vouchers quickly, and this likely caused the small businesses that offered deals to become overwhelmed. It is not surprising, given our findings, that both Groupon and the Korean daily deal site abandoned the use of the tipping point and modified redemption rules.

In 2008, a Korean online fashion distributor, Accommate, is founded in China and it tapped into the Chinese online market with the alliance of Taobao, online C2C(consumer-to-consumer) commerce platform of Alibaba. Taobao offered free listings to attract more sellers in order to compete e-bay and introduced keyword search features which Accommate made use of as an early adopter and partner. As Taobao became China’s market leader, Accommate secured the partner position to distribute Korean apparels and strengthened its partnership with Alibaba. Soon Alibaba introduced a new online brand mall called Taobao Mall or Tmall to complement the C2C platform. Accommate extended its business with Alibaba by taking charge of the Korean product sections in the online brand mall of Tmall. From the perspective of scales, China represented for Accommate a dynamic and fast-growing market, though it is so competitive. Moreover, it is a tough job

Projections of long-term carbon dioxide (CO2) emissions in the literature vary over a wide range depending on both socio-economic development and mitigation policy. Therefore, any attempt to improve the accuracy of such projections is important. We present a simple aggregate model to project the future carbon intensity of economic output, which is then used to forecast energy-related CO2 emissions through 2040 for seven countries and three regions of the world. Our projection results are compared to those from the International Energy Outlook 2013 report. We find that our projections are significantly higher than those reported in the International Energy Outlook 2013 report.

This research examines the existence of urban safety advantage versus penalty among 49 to 42 large cities. Each city is scored across four main categories of safety: digital security, health security, infrastructure safety, and personal safety for the year of 2015. Total group of cities is divided into three subgroups of cities with different population size. Multivariate cross sectional regression of five safety indexes as function of population size has generated positive population coefficients indicating that urban safety advantage dose exist. The scale advantage becomes very pronounced as the analysis moves from the total group to the subgroup of top 6 and 18 most populated cities. Implications from these findings will be discussed.

Do countries with the highest air pollution in 2000 experience further deterioration during the next 14 years, while countries with the lowest air pollution experience further improvement during the same period? To answer this question, three methods of comparative analysis were used for the four groups of 30 countries each as well as to their subgroups of countries. Results from the research indicate that the 15 countries with the highest air pollution experienced further deterioration in their averaged air pollution, whereas the next 15 countries with still high air pollution experienced a moderate improvement. The 30 countries with the lowest air pollution experienced a rapid improvement. In addition, all eight subgroups of countries displayed statistically valid gamma convergence indicating that catch-up effect has taken place. Policy implications from these findings will be discussed.

The high level of urbanisation in the USA may have led to the generation of very high emissions of carbon dioxide (CO2) in cities. This study examines the scaling relationship of CO2 emission as a function of population size across a large number of US cities, conducting scaling analyses on multiple subgroups of cities with varying population sizes. The findings indicate that while superlinear or linear scaling can be applied to the total sample of cities, these changes to sublinear scaling for the subgroups of more populous cities. In addition, this study finds that the tipping point toward said sublinear scaling lies at the subgroup of the 75 most populous cities. The implications of the findings will be discussed.

Past projections of wind-powered electricity by several organizations including Energy Information Administration have been too conservative in the past. We generate alternative projections for China, the U.S., India, Canada, Japan and South Korea and compare them against the projections made by EIA for the years of 2020, 2030, and 2040. Although our projections are much higher than the EIA’s for all five countries with the exception of Japan, our projections are supported by several projections made by other organizations including the respective government sources. Implications from our findings will be discussed as well.