Understanding Emojis for Financial Sentiment Analysis

2023 | International Conference on Information Systems | Citations: 0

Authors: Chen, Siyi; Xing, Frank

Abstract: ... Expand

Abstract: Collapse

Topics: application programming interface social media website cryptocurrency Python

Methods: sentiment analysis machine learning BERT deep learning experiment

Fraud Detection in Mobile Payment Systems using an XGBoost-based Framework

2023 | Information Systems Frontiers | Citations: 0

Authors: Hajek, Petr; Abedin, Mohammad Zoynul; Sivarajah, Uthayasankar

Abstract: Mobile payment systems are becoming more popular due to the increase in the numb ... Expand

Abstract: Mobile payment systems are becoming more popular due to the increase in the number of smartphones, which, in turn, attracts the interest of fraudsters. Extant research has therefore developed various fraud detection methods using supervised machine learning. However, sufficient labeled data are rarely available and their detection performance is negatively affected by the extreme class imbalance in financial fraud data. The purpose of this study is to propose an XGBoost-based fraud detection framework while considering the financial consequences of fraud detection systems. The framework was empirically validated on a large dataset of more than 6 million mobile transactions. To demonstrate the effectiveness of the proposed framework, we conducted a comparative evaluation of existing machine learning methods designed for modeling imbalanced data and outlier detection. The results suggest that in terms of standard classification measures, the proposed semi-supervised ensemble model integrating multiple unsupervised outlier detection algorithms and an XGBoost classifier achieves the best results, while the highest cost savings can be achieved by combining random under-sampling and XGBoost methods. This study has therefore financial implications for organizations to make appropriate decisions regarding the implementation of effective fraud detection systems. Collapse

Topics: fraud detection mobile system mobile payment electronic payment Python

Methods: XGBoost machine learning anomaly detection supervised learning experiment

Responsible Artificial Intelligence in Healthcare: Predicting and Preventing Insurance Claim Denials for Economic and Social Wellbeing

2023 | Information Systems Frontiers | Citations: 1

Authors: Johnson, Marina; Albizri, Abdullah; Harfouche, Antoine

Abstract: It is estimated that one out of seven health insurance claims is rejected in the ... Expand

Abstract: It is estimated that one out of seven health insurance claims is rejected in the US; hospitals across the country lose approximately $262 billion annually due to denied claims. This widespread problem causes huge cash-flow issues and overburdens patients. Thus, preventing claim denials before claims are submitted to insurers improves profitability, accelerates the revenue cycle, and supports patients’ wellbeing. This study utilizes Design Science Research (DSR) paradigm and develops a Responsible Artificial Intelligence (RAI) solution helping hospital administrators identify potentially denied claims. Guided by five principles, this framework utilizes six AI algorithms – classified as white-box and glass-box – and employs cross-validation to tune hyperparameters and determine the best model. The results show that a white-box algorithm (AdaBoost) model yields an AUC rate of 0.83, outperforming all other models. This research’s primary implications are to (1) help providers reduce operational costs and increase the efficiency of insurance claim processes (2) help patients focus on their recovery instead of dealing with appealing claims. Collapse

Topics: artificial intelligence ADA missing data partial dependence plot responsible artificial intelligence

Methods: machine learning computational algorithm experiment support vector machine design artifact

Reputation Spillover from Agencies on Online Platforms: Evidence from the Entertainment Industry

2023 | Management Information Systems Quarterly | Citations: 0

Authors: Kokkodis, Marios; Adamopoulos, Panagiotis; Ransbotham, Sam

Abstract: Digital markets have proliferated in recent years, overcoming many market ineff ... Expand

Abstract: Digital markets have proliferated in recent years, overcoming many market inefficiencies by facilitating direct interactions between consumers and creators. Thanks to this disintermediation, consumers now have access to a vast number of alternatives, while creators can efficiently reach huge markets. However, the success of digital markets has created a concomitant challenge for creators: differentiation. In crowded markets, agencies (e.g., publishing companies in books, freelance agencies in online labor markets, independent labels in music) can differentiate creators by signaling product quality. But how do agencies' reputations affect product success for creators? Can some agencies do more harm than good? To investigate these research questions, we theorize how variation in creator and agency reputation leads to asymmetric and heterogeneous effects, namely that (1) more reputable agencies have a stronger positive effect on less reputable creators than they have on more reputable creators, and (2) less reputable agencies hurt more reputable creators more than they hurt less reputable ones. Analyses of more than one million observations from two digital markets provide empirical support for these theory-driven arguments. The findings have design implications for markets and contribute to our understanding of how agencies, depending on creator reputation, can either benefit or hurt product success.Marios Kokkodis is a machine learning scientist working on personalization, recommender systems, and markets. His work appears in major information systems, data mining, and multidisciplinary outlets. He has received multiple awards and nominations, including the INFORMS ISS Gordon B. Davis Young Scholar Award, the INFORMS ISS Nunamaker-Chen dissertation award, the NYU Stern Harold W. MacDowell award, and the LinkedIn economic graph challenge award. Collapse

Topics: electronic market social media Google+ digital platform marketing management

Methods: theory development econometric modeling descriptive statistic gradient boosting machine learning

Theories: cognitive theory expectation disconfirmation theory decision theory contrast theory assimilation theory of learning

Driving risk prevention in usage-based insurance services based on interpretable machine learning and telematics data

2023 | Decision Support Systems | Citations: 0

Authors: Li, Hong-Jie; Luo, Xing-Gang; Zhang, Zhong-Liang; Jiang, Wei; Huang, Shen-Wei

Abstract: Usage-based insurance (UBI) adjusts premiums based on an individual policyholder ... Expand

Abstract: Usage-based insurance (UBI) adjusts premiums based on an individual policyholder’s dynamic risk evaluation, incentivizing policyholders to maintain safe driving behavior in pursuit of a lower insurance premium. Although post-trip interventions can improve driving behavior significantly, most interventions do not provide risk mitigation strategies tailored to long-term driving behavior. This study proposes a novel approach for driving risk prevention in UBI services, providing risk analysis services and tailoring driving suggestions for policyholders. Advanced prediction models and model interpretation methods were adopted to assess and analyze personal driving risks, respectively. A multi-objective counterfactual model was developed to generate risk mitigation strategies for policyholders. The significance of driving behavior in risk mitigation and its actionability were assessed to provide realistic and actionable suggestions for policyholders. The case study, based on a real dataset from an insurance company in China, showed that the proposed approach can be directly embedded in the existing UBI service framework to provide personalized feedback and services to policyholders. Collapse

Topics: shapley additive explanation decision support system Python decision making logistics management

Methods: experiment computational algorithm machine learning XGBoost factor analysis

Reviewing from a Distance: Uncovering Asymmetric Moderations of Spatial and Temporal Distance between Sentiment Negativity and Rating

2023 | Management Information Systems Quarterly | Citations: 0

Authors: Neumann, Jürgen; Gutt, Dominik; Kundisch, Dennis

Abstract: Drawing on construal level theory, prior literature has found a positivity bias ... Expand

Abstract: Drawing on construal level theory, prior literature has found a positivity bias in online ratings when consumers evaluate an experience from a psychological distance, whether spatial or temporal. Selfdistancing theory posits that psychological distance enables individuals to reflect on psychologically distant negative experiences more genuinely, in a less biased way. This raises the question of whether the positivity bias in ratings due to psychological distance persists for negative experiences. To address this question, we collected data from a large review platform that enables the identification of reviewers' spatial and temporal distance. The negativity of an experience was operationalized via review text sentiment. We introduced spatial and temporal distance as moderators between sentiment negativity and ratings and found a negative moderation by spatial distance and a positive moderation by temporal distance. Our findings indicate that the relationship between sentiment negativity and rating grows stronger under spatial distance and gets weaker under temporal distance. Text mining confirmed selfdistancing as the driver behind the spatial moderation and construal levels as the driver behind the temporal moderation. We attribute the asymmetric moderations to differences in the tangibility of spatial distance (more tangible) and temporal distance (less tangible). These results improve our understanding of reviewing behavior and can help platforms de-bias ratings. Collapse

Topics: online review decision making user-generated content website media richness

Methods: product review theory development robustness check sentiment analysis descriptive statistic

Theories: linguistic category model construal level theory media richness theory

Peer-to-Peer (P2P) Lending Risk Management: Assessing Credit Risk on Social Lending Platforms Using Textual Factors

2023 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Siering, Michael

Abstract: Peer-to-peer (P2P) lending platforms offer Internet users the possibility to bor ... Expand

Abstract: Peer-to-peer (P2P) lending platforms offer Internet users the possibility to borrow money from peers without the intervention of traditional financial institutions. Due to the anonymity on such social lending platforms, determining the creditworthiness of borrowers is of high importance. Beyond the disclosure of traditional financial variables that enable risk assessment, peer-to-peer lending platforms offer the opportunity to reveal additional information on the loan purpose. We investigate whether this self-disclosed information is used to show reliability and to outline creditworthiness of platform participants. We analyze more than 70,000 loans funded at a leading social lending platform. We show that linguistic and content-based factors help to explain a loan's probability of default and that content-based factors are more important than linguistic variables. Surprisingly, not every information provided by borrowers underlines creditworthiness. Instead, certain aspects rather indicate a higher probability of default. Our study provides important insights on information disclosure in the context of peer-to-peer lending, shows how to increase performance in credit scoring and is highly relevant for the stakeholders on social lending platforms. Collapse

Topics: crowdsourcing peer-to-peer lending LendingClub risk management logistics management

Methods: qualitative content analysis machine learning regression analysis method descriptive statistic logistic regression

Theories: signaling theory

Framework for the Classification of Imbalanced Structured Data Using Under-sampling and Convolutional Neural Network

2022 | Information Systems Frontiers | Citations: 0

Authors: Lee, Yoon Sang; Bang, Chulhwan Chris

Abstract: Among machine learning techniques, classification techniques are useful for vari ... Expand

Abstract: Among machine learning techniques, classification techniques are useful for various business applications, but classification algorithms perform poorly with imbalanced data. In this study, we propose a classification technique with improved binary classification performance on both the minority and majority classes of imbalanced structured data. The proposed framework is composed of three steps. In the first step, a balanced training set is created via under-sampling. Then, each example is converted into an image depicting a line graph. In the last step, a Convolutional Neural Network (CNN) is trained using the images. In the experiments, we selected six datasets from the UCI Repository and applied the proposed framework to them. The proposed model achieved the best receiver operating characteristic (ROC) curve and Balanced Accuracy (BA) on all the datasets and five datasets, respectively. This demonstrates that the combination of under-sampling and CNNs is a viable approach for imbalanced structure data classification. Collapse

Topics: marketing management Python systems implementation packaged software

Methods: machine learning decision tree classification AdaBoost experiment binary classification

Show Me Your Claims and I'll Tell You Your Offenses: Machine Learning-Based Decision Support for Fraud Detection on Medical Claim Data

2022 | HICSS | Citations: 0

Authors: Matschak, Tizian; Prinz, Christoph; Rampold, Florian; Trang, Simon

Abstract: Health insurance claim fraud is a serious issue for the healthcare industry as i ... Expand

Abstract: Health insurance claim fraud is a serious issue for the healthcare industry as it drives up costs and inefficiency. Therefore, claim fraud must be effectively detected to provide economical and high-quality healthcare. In practice, however, fraud detection is mainly performed by domain experts resulting in significant cost and resource consumption. This paper presents a novel Convolutional Neural Network-based fraud detection approach that was developed, implemented, and evaluated on Medicare Part B records. The model aids manual fraud detection by classifying potential types of fraud, which can then be specifically analyzed. Our model is the first of its kind for Medicare data, yields an AUC of 0.7 for selected fraud types and provides an applicable method for medical claim fraud detection. Collapse

Topics: fraud detection open source decision support knowledge base

Methods: machine learning literature study decision tree classification mixed method feature engineering

Buyers’ Strategic Behavior in B2B Multichannel Auction Markets: When an Online Posted Price Channel Is Incorporated into a Dutch Auction System

2022 | Information Systems Research | Citations: 0

Authors: Truong, May; Gupta, Alok; Ketter, Wolfgang; van Heck, Eric

Abstract: Firms are increasingly adopting different sales channels to reach new potential ... Expand

Abstract: Firms are increasingly adopting different sales channels to reach new potential buyers. Yet, extant research has mainly focused on business-to-customer (B2C) online and offline posted price channels. Business-to-business (B2B) multichannel and, especially, systems with multiple pricing mechanisms are largely underexplored. This paper investigates the strategic behaviors of B2B buyers in a unique system where an online posted price channel is incorporated into a Dutch auction market sequentially. We follow buyers’ purchasing paths and examine conditions under which B2B auction buyers will use the online posted price channel. We incorporate learning and experience and demonstrate how buyers’ behaviors evolve. We investigate an emerging group of buyers who use different price mechanisms and their surplus extraction activities. We further explore how the market flow changes when posted prices are incorporated. Our results, using an extensive data set from the world’s largest flower market, highlight the importance of quantity demand, product diversity, and experience in explaining the choice of the new posted price channel. We find a significantly higher average loss of surplus at the product level for multichannel buyers than for single channel buyers and a reduction in the number of small orders in the auction channel. Subsequently, theoretical and managerial implications for B2B multichannel markets and market design are discussed. History: Robert Day, Senior Editor; Xinxin Li, Associate Editor. This paper has been accepted for the Information Systems Research Special Section on Market Design and Analytics. Collapse

Topics: business-to-business business-to-consumer price management online auction electronic commerce

Methods: robustness check scale validation descriptive statistic cluster analysis chi squared test