Domain-independent real-time service provisioning in digital platforms: Featuring bundling and customer time-preference

2023 | Decision Support Systems | Citations: 0

Authors: Mukherjee, Anik; Rangaraja P., Sundarraj; Vander Meer, Debra; Dutta, Kaushik

Abstract: Digital platforms have emerged as an important technology underpinning the new e ... Expand

Abstract: Digital platforms have emerged as an important technology underpinning the new economy. A key problem in such platforms concerns provisioning decisions for customer-service requests, in order to maximize the pro­ vider’s revenue subject to resource availability. Provisioning is important to meet customer needs, and in turn, for customer retention. The service provisioning problem, in addition to being NP-hard, is further compounded by the fact that the timeframe to make provisioning decisions is short, and more importantly, varies considerably across different platform domains. Digital-platform customers often request bundled services and are sensitive to when tasks get completed. We describe a behavioral-economic model of this problem and develop a domainindependent solution featuring a judicious combination of optimization-backed rule-based approach. Through extensive computational results, we show how our approach has the potential for generalizability and tailor­ ability across different domains in a seamless manner, thus offering the advantages proposed in the softwarereusability literature. Implications to platform sustainability, reusable platform solutions and to practical ap­ proaches to real-time platforms are also provided. Collapse

Topics: digital platform decision making cloud computing ride sharing platform smartphone

Methods: experiment computational algorithm knapsack problem association rule learning simulation

Theories: behavioral economics

Will they take this offer? A machine learning price elasticity model for predicting upselling acceptance of premium airline seating

2023 | Information & Management | Citations: 0

Authors: Thirumuruganathan, Saravanan; Emadi, Noora Al; Jung, Soon-gyo; Salminen, Joni; Robillos, Dianne Ramirez; Jansen, Bernard J.

Abstract: Employing customer information from one of the world’s largest airline companies ... Expand

Abstract: Employing customer information from one of the world’s largest airline companies, we develop a price elasticity model (PREM) using machine learning to identify customers likely to purchase an upgrade offer from economy to premium class and predict a customer’s acceptable price range. A simulation of 64.3 million flight bookings and 14.1 million email offers over three years mirroring actual data indicates that PREM implementation results in approximately 1.12 million (7.94%) fewer non-relevant customer email messages, a predicted increase of 72,200 (37.2%) offers accepted, and an estimated $72.2 million (37.2%) of increased revenue. Our results illustrate the potential of automated pricing information and targeting marketing messages for upselling acceptance. We also identified three customer segments: (1) Never Upgrades are those who never take the upgrade offer, (2) Upgrade Lovers are those who generally upgrade, and (3) Upgrade Lover Lookalikes have no historical record but fit the profile of those that tend to upgrade. We discuss the implications for airline companies and related travel and tourism industries. Collapse

Topics: electronic mail marketing management price management customer satisfaction application programming interface

Methods: autoencoder data transformation machine learning experiment principal component analysis

Theories: optimization theory decision theory

Diversity Preference-Aware Link Recommendation for Online Social Networks

2023 | Information Systems Research | Citations: 0

Authors: Yin, Kexin; Fang, Xiao; Chen, Bintong; Sheng, Olivia R. Liu

Abstract: Link recommendation, which recommends links to connect unlinked online social ne ... Expand

Abstract: Link recommendation, which recommends links to connect unlinked online social network users, is a fundamental social network analytics problem with ample business implications. Existing link recommendation methods tend to recommend similar friends to a user but overlook the user’s diversity preference, although social psychology theories suggest the criticality of diversity preference to link recommendation performance. In recommender systems, a field related to link recommendation, a number of diversification methods have been proposed to improve the diversity of recommended items. Nevertheless, diversity preference is distinct from diversity studied by diversification methods. To address these research gaps, we define and operationalize the concept of diversity preference for link recommendation and propose a new link recommendation problem: the diversity preference-aware link recommendation problem. We then analyze key properties of the new link recommendation problem and develop a novel link recommendation method to solve the problem. Using two large-scale online social network data sets, we conduct extensive empirical evaluations to demonstrate the superior performance of our method over representative diversification methods adapted for link recommendation and state-of-the-art link recommendation methods. History: Eric Zheng, Senior Editor; Zhengrui Jiang, Associate Editor. Supplemental Material: The online appendices are available at https://doi.org/10.1287/isre.2022.1174. Collapse

Topics: recommender system social network resource allocation Google+

Methods: machine learning optimization model computational algorithm experiment integer programming

Bidder Support in Multi-item Multi-unit Continuous Combinatorial Auctions: A Unifying Theoretical Framework

2022 | Information Systems Research | Citations: 0

Authors: Adomavicius, Gediminas; Gupta, Alok; Yang, Mochen

Abstract: Despite known advantages of combinatorial auctions, wide adoption of this alloca ... Expand

Abstract: Despite known advantages of combinatorial auctions, wide adoption of this allocation mechanism, especially in consumer-oriented marketplaces, is limited partially by the lack of effective bidder support information that can assist bidders to make bidding decisions. In this paper, we study the bidder support problem for general multi-item multi-unit (MIMU) combinatorial auctions, where multiple heterogeneous items are being auctioned and multiple homogeneous units are available for each item. Specifically, we consider continuous MIMU auctions, which impose minimal restrictions on bidding activities, thereby reducing the complexity of participation. Two prevalent bidding languages: OR bidding and XOR bidding, are discussed separately. For MIMU auctions with XOR bids, we derive theoretical results to calculate important bidder support metrics. We further demonstrate that bidder support results for MIMU auctions with OR bids can be derived directly from those with XOR bids, by viewing OR bids as XOR bids with each bid submitted by a unique bidder. Consequently, we establish MIMU auctions with XOR bids as the most general case, and unify the theoretical insights on bidder support problem for different bidding languages as well as different special cases of general MIMU auctions, namely single-item multi-unit (SIMU) auctions and multi-item single-unit (MISU) auctions. The derived theoretical results lead to algorithmic procedures that are capable of providing bidder support information efficiently in practice, and that outperform the commonly used integer programming approach. Theoretical insights of the general MIMU auctions also extend to auctions with additional bidding constraints, including batch-based combinatorial auctions, hierarchical combinatorial auctions, and combinatorial reverse auctions. History: This paper has been accepted for the Information Systems Research Special Section on Market Design and Analytics. Supplemental Material: The online appendices are available at https://doi.org/10.1287/isre.2021.1068. Collapse

Topics: procurement management Linux computer hardware programming language accounting

Methods: simulation computational algorithm descriptive statistic integer programming dynamic programming

Theories: game theory

An Integer Programming Approach to Solving the Inverse Graph Model for Conflict Resolution with Two Decision Makers

2022 | Group Decision and Negotiation | Citations: 0

Authors: Han, Yu; Xu, Haiyan; Fang, Liping; Hipel, Keith W.

Abstract: Operational algorithms for solving the inverse problem for the graph model for c ... Expand

Abstract: Operational algorithms for solving the inverse problem for the graph model for conflict resolution are presented for the case of two decision makers (DMs) under a variety of solution concepts, including Nash stability (Nash), general metarationality (GMR), symmetric metarationality (SMR), and sequential stability (SEQ). The algorithms based on integer programming enable a DM, an analyst, or a mediator to obtain all of the preferences required to make a specified state to be an equilibrium or resolution. For the cases of Nash, GMR, and SMR, the respective inverse algorithm for the focal DM is formulated as a 0–1 integer linear programming problem even when both DMs’ preferences are unknown. For the situation of SEQ, when both DMs’ preferences are unknown, the focal DM’s algorithm is a 0–1 integer nonlinear programming problem while, under the condition that the opponent’s preferences are known, the focal DM’s 0–1 integer programming problem is linear. The usefulness of the algorithms developed is demonstrated by applying them to an illustrative dispute. Collapse

Methods: integer programming computational algorithm linear optimization integer linear programming

Theories: prisoner's dilemma

Modifying Transactional Databases to Hide Sensitive Association Rules

2022 | Information Systems Research | Citations: 0

Authors: Menon, Syam; Ghoshal, Abhijeet; Sarkar, Sumit

Abstract: Firms have been sharing transactional data with business partners ever since ele ... Expand

Abstract: Firms have been sharing transactional data with business partners ever since electronic data interchange was introduced to the retail industry in the 1980s. The potential benefits of data sharing notwithstanding, there has been continued reluctance on the part of data owners to share their data, for fear of sensitive information potentially making its way to competitors. Approaches that can help hide sensitive information could alleviate such concerns and increase the number of firms that are willing to share. Sensitive information in transactional databases often manifests itself in the form of association rules. Association rules can be concealed by altering transactions such that these sensitive rules stay hidden when the data are mined. The problem of hiding sensitive association rules is NP-hard, and to date, it has only been addressed via heuristic approaches. In this paper, we introduce a nonlinear integer formulation to hide sensitive association rules while maximizing the accuracy of the altered database. We then separate it into two problems: the sanitization problem, which hides sensitive association rules from a specific transaction while altering the transaction as minimally as possible, and the accuracy maximization problem, which maximizes the accuracy of the altered database, given a solution to the sanitization problem. We show how the sanitization problem can be represented as an integer program and propose a heuristic based on intuition from this formulation to solve it. Next, we formulate the accuracy maximization problem as a nonlinear integer program, show how it can be linearized, and derive various results that help reduce the size of the problem to be solved. Computational experiments are conducted on real and synthetic data sets, the largest of which has 100 million transactions. Our results show that although the nonlinear integer formulations are not practical, the linearizations and problem-reduction steps make a significant impact on solvability and solution time. We also find that there are substantial gains to be realized vis-à-vis existing approaches in terms of both solution quality and solution time and that hiding the rules directly (rather than by hiding the associated itemsets) can result in significantly fewer transactions being sanitized. Collapse

Topics: database system accounting Microsoft Windows data quality data governance

Methods: experiment integer programming computational algorithm literature study

Personalized Recommendation for Balancing Content Generation and Usage on Two-Sided Entertainment Platforms

2022 | International Conference on Information Systems | Citations: 0

Authors: Zhang, Hao; Guo, Zhiling; Wang, Mingzheng

Abstract: Online entertainment platforms such as Youtube host a vast amount of user-gener ... Expand

Abstract: Online entertainment platforms such as Youtube host a vast amount of user-generated content (UGC). The unique feature of two-sided UGC entertainment platforms is that creators' content generation and users' content usage can influence each other. However, traditional recommender systems often emphasize content usage but ignore content generation, leading to a misalignment between these two goals. To address the challenge, this paper proposes a prescriptive uplift framework to balance content generation and usage through personalized recommendations. Specifically, we first predict the heterogeneous treatment effects (HTEs) of recommended contents on creators' content generation and users' content usage, then consider these two predicted HTEs simultaneously in an optimization model to determine the recommended contents for each user. Using a large-scale real-world dataset, we demonstrate that the proposed recommendation method better balances content generation and usage and brings a 42% increase in participants' activity compared to existing benchmark methods. Collapse

Topics: recommender system cloud computing digital service music streaming service information system use

Methods: optimization model hidden Markov model regression analysis method machine learning causality analysis

Clustering temporal disease networks to assist clinical decision support systems in visual analytics of comorbidity progression

2021 | Decision Support Systems | Citations: 3

Authors: Lu, Yajun; Chen, Suhao; Miao, Zhuqi; Delen, Dursun; Gin, Andrew

Abstract: Detection and characterization of comorbidity, the presence of more than one dis ... Expand

Abstract: Detection and characterization of comorbidity, the presence of more than one distinct disorder or illness concurrently occurring among a specific cohort of patients, is an invaluable decision aid and a prominent challenge in healthcare research and practice. The aim of this paper is to design a novel visual analytics system that can support efficient pattern detection and intuitive visualization of comorbidity progression modeled via temporal disease networks (TDNs). In the underlying system, we proposed two new clustering tech­ nologies—temporal clustering and disease clustering to detect the time of notable progression changes and simplify the visualization of TDNs. Through two case studies on Clostridioides Difficile and stroke, we demon­ strate that the proposed system is able to provide evidence-based and visual insights regarding comorbidity progression effectively for clinical decision support. Collapse

Topics: health information system Python data warehouse decision support systems design

Methods: cluster analysis computational algorithm case study literature study structured literature research

Time-preference-based on-spot bundled cloud-service provisioning

2021 | Decision Support Systems | Citations: 0

Authors: Mukherjee, Anik; Sundarraj, Rangaraja P.; Dutta, Kaushik

Abstract: The cloud computing spot instance is one offering that vendors are leveraging to ... Expand

Abstract: The cloud computing spot instance is one offering that vendors are leveraging to provide differentiated service to an expanding pay-per-use computing market. Spot instances have cost advantages, albeit at a trade-off of in­ terruptions that can occur when the user’s bid price falls below the spot price. The interruptions are often exacerbated since customers often require resources in bundles. For these reasons, customers might have to wait for a long time before their jobs are completed. In this paper, we propose a behavioral-economic model in the form of time-preference-based bids, wherein users are willing to use and bid for services at other times if the vendor cannot provide the resources at the preferred time. Given such bids, we consider the problem of provi­ sioning for such service requests. We develop a time-preference-based optimization model. Since the optimiza­ tion model is NP-Hard, we develop rule-based genetic algorithms. We have obtained very encouraging results with respect to standard commercial solver as a benchmark. In turn, our results provide evidence for the viability of our approach for online service-provisioning problems. Collapse

Topics: cloud computing cloud computing provider packaged software virtualization technology digital platform

Methods: computational algorithm experiment optimization model literature study knapsack problem

Theories: behavioral economics

A Graph-Based Ant Algorithm for the Winner Determination Problem in Combinatorial Auctions

2021 | Information Systems Research | Citations: 0

Authors: Ray, Abhishek; Ventresca, Mario; Kannan, Karthik

Abstract: Combinatorial auctions (CAs) are used to allocate bundles of items among interes ... Expand

Abstract: Combinatorial auctions (CAs) are used to allocate bundles of items among interested bidders. However, to resolve bidder preference elicitation problems, CAs are conducted iteratively. Winner determination is a key bottleneck that restricted the widespread adoption of such iterative CAs. Time bounded winner determination is further complicated by the increased variety and velocity of bids each round, and so regular solvers such as IBM CPLEX and A Mathematical Programming Language have been demonstrably ineffective in determining winners within a short duration. In response, heuristic-based methods have enjoyed an increase in applicability. We add to the growing body of work in heuristic-based winner determination by proposing an ant colony–based anytime algorithm (traveling ants for combinatorial auctions, or TrACA) that produces optimal or near-optimal winner determination results within specified time. Experiments are performed on 94 open test instances that display a variety in bid-bundle composition. We show and compare the performance of TrACA to 20 most recent state-of-the-art heuristics and two of the best exact algorithms (CPLEX and Max W Clique). Compared with 12 of 20 heuristics that employ classic search (e.g., stochastic local search, tabu search) or genetic or memetic algorithms, TrACA does significantly better on all instances. On the other eight heuristics that employ advanced hybrid search methodologies (e.g., hyperheuristics, biased random keys), TrACA performs significantly better on the harder test instances. Additionally, we show that TrACA resolves the trade-off between time and accuracy by achieving optimal solutions on 76 out of 94 instances, in at most one-sixth the time taken by exact algorithms. Finally, we analyze the TrACA search process and highlight factors that make TrACA suitable for solving hard auction instances within a prespecified (short) duration. Collapse

Topics: source code random-access memory Microsoft Windows virtualization technology cloud computing

Methods: computational algorithm experiment nonparametric test Mann Whitney U test optimization model