Robert S. Garfinkel

Empirical Analysis of the Impact of Recommender Systems on Sales

Online retailers are increasingly using information technologies to provide value-added services to customers. Prominent examples of these services are online recommender systems and consumer feedback mechanisms, both of which serve to reduce consumer search costs and uncertainty associated with the purchase of unfamiliar products. The central question we address is how recommender systems affect sales. We take into consideration the interaction among recommendations, sales, and price. We then develop a robust empirical model that incorporates the indirect effect of recommendations on sales through retailer pricing, potential simultaneity between sales and recommendations, and a comprehensive measure of the strength of recommendations. Applying the model to a panel data set collected from two online retailers, we found that the strength of recommendations has a positive effect on sales. Moreover, this effect is moderated by the recency effect, where more recently released recommended items positively affect the cross-selling efforts of sellers. We also show that recommender systems help to reinforce the long-tail phenomenon of electronic commerce, and obscure recommendations positively affect cross-selling. We also found a positive effect of recommendations on prices. These results suggest that recommendations not only improve sales but they also provide added flexibility to retailers to adjust their prices. A comparative analysis reveals that recommendations have a higher effect on sales than does consumer feedback. Our empirical results show that providing value-added services, such as digital word of mouth and recommendations, allows retailers to charge higher prices while at the same time increasing demand by providing more information regarding the quality and match of products.

Finite dominating sets for network location problems

A research theme involving location on networks, since its inception, has been the identification of a finite dominating set (FDS), or a finite set of points to which an optimal solution must belong. We attempt to unify and generalize results of this sort. We survey the literature and then prove some theorems that subsume most previous results and that are, at the same time, more general than previous results. The paper is aimed primarily at investigators who wish to know whether an FDS exists for a specific problem.

A Market Design for Grid Computing

Grid computing uses software to integrate computing resources, such as CPU cycles, storage, network bandwidth, and even applications, across a distributed and heterogeneous set of networked computers. It is now widely deployed by organizations and provides seamless temporary processing-capacity expansion to handle peak-period demand on e-commerce servers, distributed gaming, and content storage and distribution. We develop a market-based resource-allocation model that adds an economic layer to the current approach of treating resource allocation as primarily a scheduling issue. We design a value-elicitation and allocation scheme that provides the economic incentives for buyers and sellers of computing resources to exchange assets. We formulate the problem as a combinatorial call auction and present a portfolio of three solution approaches that trade off economic properties, such as allocative efficiency, incentive compatibility, and fairness in allocation, with computational efficiency. The first of these is an efficient solution that maximizes social welfare and yields incentive-compatible Vickrey-Clarke-Groves prices, but requires solving multiple instances of an NP-hard problem. For markets where having a commodity price is critical, we show how the addition of fairness constraints to the efficient model can somewhat reduce the computational burden and yet preserve incentive compatibility. Finally, for markets that require real-time fast solution techniques, we propose a time-sensitive fair Grid (tsfGRID) heuristic that relaxes the maximal allocation requirement of the welfare-maximizing fair solution. Its solution is not guaranteed to be incentive-compatible, but the heuristic is designed to be fast, maintain fairness in allocations, and yield commodity prices. Notably, while incentive compatibility is not guaranteed by tsfGRID, computational results comparing it with the efficient solution technique indicate that there are no significant differences in the expected-revenue and operational-allocative characteristics.

Technical Note—An Improved Algorithm for the Bottleneck Assignment Problem

Gross has proposed an algorithm for the bottleneck assignment problem. This note reports a test of it against a “threshold” algorithm, and finds that the latter is superior computationally.

OR Practice---Efficient Short-Term Allocation and Reallocation of Patients to Floors of a Hospital During Demand Surges

Many hospitals face the problem of insufficient capacity to meet demand for inpatient beds, especially during demand surges. This results in quality degradation of patient care due to large delays from admission time to the hospital until arrival at a floor. In addition, there is loss of revenue because of the inability to provide service to potential patients. A solution to the problem is to proactively transfer patients between floors in anticipation of a demand surge. Optimal reallocation poses an extraordinarily complex problem that can be modeled as a finite-horizon Markov decision process. Based on the optimization model, a decision-support system has been developed and implemented at Windham Hospital in Willimantic, Connecticut. Projections from an initial trial period indicate very significant financial gains of about 1% of their total revenue, with no negative impact on any standard quality of care or staffing effectiveness indicators. In addition, the hospital showed a marked improvement in quality of care because of a resulting decrease of almost 50% in the average time that an admitted patient has to wait from admission until being transferred to a floor.

Privacy Protection of Binary Confidential Data Against Deterministic, Stochastic, and Insider Threat

A practical model and an associated method are developed for providing consistent, deterministically correct responses to ad-hoc queries to a database containing a field of binary confidential data. COUNT queries, i.e., the number of selected subjects whose confidential datum is positive, are to be answered. Exact answers may allow users to determine an individuals confidential information. Instead, the proposed technique gives responses in the form of a number plus a guarantee so that the user can determine an interval that is sure to contain the exact answer. At the same time, the method is also able to provide both deterministic and stochastic protection of the confidential data to the subjects of the database. Insider threat is defined precisely and a simple option for defense against it is given. Computational results on a simulated database are very encouraging in that most queries are answered with tight intervals, and that the quality of the responses improves with the number of subjects identified by the query. Thus the results are very appropriate for the very large databases prevalent in business and governmental organizations. The technique is very efficient in terms of both time and storage requirements, and is readily scalable and implementable.

Optimal Imputation of Erroneous Data: Categorical Data, General Edits

Responses to surveys often contain large amounts of incorrect information. One option for dealing with the problem is to revise those erroneous responses that can be detected. Fellegi and Holt developed a model in which a response is modified to pass a set of edits with as little change as possible. The model is called Minimum Weighted Fields to Impute MWFI and is NP-hard for categorical data and general edits. We develop two algorithms for MWFI, based on set covering, and present computational experience.

The Bottleneck Traveling Salesman Problem: Algorithms and Probabilistic Analysis

The bottleneck traveling salesman problem seeks to minimize the maximum length arc over all Hamlltonlan cycles in a graph A probabdlStlC analysis is presented for random problems It is shown that the optimal ob.lectwe value can be closely approximated by a beta function Finally, effective solution techmques are developed and computational experience is reported

Confidentiality via Camouflage: The CVC Approach to Disclosure Limitation When Answering Queries to Databases

A practical method is presented for giving unlimited, deterministically correct, numerical responses to ad-hoc queries to an online database, while not compromising confidential numerical data. The method is appropriate for any size database, and no assumptions are needed about the statistical distribution of the confidential data. Responses are in the form of a number plus a guarantee, so the user can determine an interval that is sure to contain the exact answer. Virtually any imaginable query type can be answered, and in the absence of insider information, collusion among the users presents no problem. Experimental analysis supports the practical viability of the proposed method.

Shopbot 2.0: Integrating recommendations and promotions with comparison shopping

Recommender systems have been used by online retailers along with various promotions to attract customers. They are often in the form of a single item (best bet) along with a choice set. The majority of choice set recommendations are made based on collaborative filtering algorithms that recommend highly related items. However, we observe that very often best bets suggested by retailers are not based strictly on relatedness, since they are not members of the choice set. We found that the probability of this occurring is positively related to the popularity of the original requested item (base item). We also show that, even when best bets are closely related to base items, there are alternate options for the best bet that are still highly related, and at the same time can integrate with existing promotions to be more appealing to price sensitive customers. We argue that shopbots are in the best position to provide such integrated service and we therefore develop an integer programming model to optimize recommendations for shopbots. This model is validated using data from two online book retailers to show that significant extra savings can be achieved by suggesting alternate best bets.