Top-k querying can significantly improve the performance of web-based business intelligence applications such as price comparison and product recommendation systems. Top-k retrieval involves finding a limited number of records in a relational database that are most similar to user-specified attribute-value pairs. This paper extends the cost-based query-mapping method for top-k retrieval by incorporating data skewness in range estimation. Experiments on real world and synthetic multi-attribute data sets show that incorporating data skewness provides a robust performance across different types of data sets, query sets, distance functions, and histograms.

Relational databases have increasingly become the basis for a wide range of applications that require efficient methods for exploratory search and retrieval. Top-k retrieval addresses this need and involves finding a limited number of records whose attribute values are the closest to those specified in a query. One of the approaches in the recent literature is query-mapping which deals with converting top-k queries into equivalent range queries that relational database management systems (RDBMSs) normally support. This approach combines the advantages of simplicity as well as practicality by avoiding the need for modifications to the query engine, or specialized data structures and indexing techniques to handle top-k queries separately. This paper reviews existing query-mapping techniques in the literature and presents a range query estimation method based on cost modeling. Experiments on real world and synthetic data sets show that the cost-based range estimation method performs at least as well as prior methods and avoids the need to calibrate workloads on specific database contents.

Finding efficient methods for supporting top-k relational queries has received significant attention in academic research. One of the approaches in the recent literature is query-mapping, in which top-k queries are mapped (translated) into equivalent range queries that relational database systems (RDBMSs) normally support. This approach combines the advantage of simplicity as well as practicality by avoiding the need for modifications to the query engine, or specialized data structures or indexing techniques to handle top-k queries separately. However, existing methods following this approach fall short of adequately modeling the problem environment and providing consistent results. In this article, the authors propose a cost-based range estimation model for the query-mapping approach. They provide a methodology for trading-off relevant query execution cost components and mapping a top-k query into a cost-optimal range query for efficient execution. Their experiments on real world and synthetic data sets show that the proposed strategy not only avoids the need to calibrate workloads on specific database contents, but also performs at least as well as prior methods.

Theoretical expectations suggest that with increasing competition online retailers would lower prices and within a given market, online retailers with higher levels of service quality would charge higher prices. Empirical investigations of online price differentiation have found reduction in price dispersion with increase in number of competitors. However, the effect of service quality on online retailers' prices has not been found to be consistent. Empirical evidence from offline markets indicates that retailers customize prices at the brand store level and that market characteristics that reflect competition are a major determinant of their pricing strategies. Together, these findings lead us to believe that a better understanding of the sources of online price dispersion can be obtained by examining interactions between retailer and market characteristics. In this paper we examine if retailer characteristics such as service quality and transaction channels impact online prices. Departing from earlier studies, we specifically seek to answer when and how, in the context of different market characteristics, do they enable the retailers to price differentiate? The market characteristics examined include number of competitors, nature of competition, and scope for differentiation in terms of both service quality and transaction channels, Given that the retailer and market level determinants are at different levels of abstractions, simple linear models are not suitable for capturing the interactions between them. We adopt a hierarchical linear modeling approach that closely reflects the levels of abstraction in the market structure and allows us to test for interactions between retailer and market characteristics. Using 13,393 price quotes for 1,880 best selling products across eight product categories, from 194 Internet retailers, we find that retailer and market characteristics interact significantly in influencing retailer price levels. The better fit between the model and the online market structure is reflected by an 25% increase in the explanation of price dispersion as compared to similar studies. Extending prior research, and consistent with theoretical expectations, we find empirical evidence for a significant positive influence of service quality on posted prices across all product categories. We observe that an increase in number of competitors induces a downward pressure in prices for all retailers, albeit at a decreasing rate. Interestingly, highly competitive markets, marked by the presence of a large number of competitors, a wider variance in the service quality of the retailers selling the product accentuates the price-premiums demanded by high service quality retailers. On the other hand, retailers with low service quality charge higher prices either when there are a large number of competitors and low variation in service quality or when there is a large variance in service quality and few competitors. Our results indicate that the posted prices of national brick-and-click retailers are sensitive to competition from other national brick-and-click retailers selling the product. Our study demonstrates that while retailer characteristics do significantly influence online prices, their impact is enhanced or diminished by the market characteristics.

An increasing number of application areas now rely on obtaining the “best matches” to a given query as opposed to exact matches sought by traditional transactions. This type of exploratory querying (also called top-k querying) can significantly improve the performance of web-based applications such as consumer reviews, price comparisons and recommendations for products/services. Due to the lack of support for specialized indexes and/or data structures in relational database management systems (RDBMSs), recent research has focused on utilizing summary statistics (histograms) maintained by RDBMSs for translating the top-k request into a traditional range query. Because the RDBMS query engines are already optimized for execution of range queries, such approach has both practical as well as efficiency advantages. In this paper, we review the strengths and weaknesses of common histogram construction techniques with regard to their structural characteristics, accuracy in approximating the true distribution of the underlying data, and implications for top-k retrieval. We also present our top-k retrieval strategy (Query-Level Optimal Cost Strategy — QLOCS) and demonstrate its “histogram-independent” performance. Based on comparative experimental and statistical analyses with the best-known histogram-based strategy in the literature, we show that QLOCS is not only more efficient but also provides more consistent performance across commonly used histogram types in RDBMSs.

Theories from Information Systems, Marketing and Economics suggest that product, retailer, and market characteristics jointly serve as determinants of online retailers' ability to price differentiate. Until now, the empirical research has focused on examining the impact of these determinants in isolation. In this paper, we extend the prior online price dispersion literature by examining the interactions among product, retailer and market characteristics. We construct a multi-level hierarchical linear model to empirically test whether market level characteristics moderate retailer characteristics in explaining price dispersion. Our analysis is based on a dataset of 13,393 price quotes for 1880 best selling products across eight product categories from 194 online retailers. The analysis indicates that service quality has a positive effect on retailer price levels. We observe that the relationship between competitive intensity in a market and retailer price levels is inverted "U" shaped. In contrast, the influence of the interaction between a retailer's service quality and market level variables on retailer price levels is "U" shaped. These findings together provide the first known empirical evidence for the existence of mixed pricing strategies among online retailers. Contrary to conventional wisdom, we find that retailers providing quality service are able to charge higher prices as the competition increases.

This paper presents the design, development and validation methodology of an agent-based computational model of the B2C electronic auction marketplace. It aims at a comprehensive understanding of the varied issues governing a B2C electronic auction, incorporating the behavior of all relevant agents such as the auctioneer, the consumer and the retailer; and the environment in which these agents operate and interact. In contrast with conventional methods, agent based modeling employs a bottom-up modeling approach where behaviors of individual agents and rules for their interaction, specified at the micro level, give rise to emergent macro level phenomenon. The development methodology should ensure that agent models are aligned with theory, current knowledge of the field and observed phenomena, and output validity of the model also needs to be ascertained. Beginning with a general introduction to agent based computational modeling, this paper formalizes this alignment and validation methodology and elaborates each step, noting the rationale and means for achieving these. The manner in which this process was used in modeling B2C auctions is then described.

Nearest neighbor querying has received the most widespread application in document and multi-media retrieval systems due to the nature of the information stored and the intuitive appeal of the request for “approximate matches”. In contrast, research on the efficient retrieval of these queries over relational database management systems (RDBMSs) is limited. Despite the increasing importance of these queries in applications such as price comparison services and product recommender systems, current RDBMSs do not natively support these queries. This dissertation proposes a Query-Level Optimal Cost Strategy (QLOCS) for estimating an optimal range query for efficient k-nearest neighbor (k-NN) retrieval over a relational database. We develop an analytical model that systematically exploits the histogram information available in RDBMS and incorporates the relevant processing costs and their trade-offs in estimating a cost-optimal range query for nearest neighbor retrieval. Experimental and computational analyses using real and synthetic databases show that the technique effectively trades competing cost factors and achieves better efficiency over existing approaches. Unlike conventional querying, k-NN querying shifts the burden of constructing the relevant range query from the user to the system. This eliminates for the user an iterative refinement of the query conditions in search of the relevant information. To this end, this research has the following distinguishing features. First, the method systematically considers the tradeoffs between the important cost factors in processing top-k queries. Second, unlike existing approaches, the method is designed to deal with individual queries rather than a pre-specified workload of queries by leveraging the statistical information (histogram) available in relational databases and undertaking cost optimization at a query level. This is particularly significant because existing approaches attempt to provide solution at an aggregate level, which requires expensive calibration of the actual dataset for a specific context. Finally, the strategy clearly separates the techniques of histogram construction from the retrieval algorithm and provides robust performance across commonly used histogram construction techniques in RDBMSs.

Financial markets data present a challenging opportunity for the learning of complex patterns not readily discernable. This paper investigates the use of genetic algorithms for the mining of financial time-series for patterns aimed at the provision of trading decision models. A simple yet flexible representation for trading rules is proposed, and issues pertaining to fitness evaluation examined. Two key issues in fitness evaluation, the design of a suitable fitness function reflecting desired trading characteristics and choice of appropriate training duration, are discussed and empirically examined. Two basic measures are also proposed for characterizing rules obtained with alternate fitness criteria.

Winner’s curse theories in case of (almost) common-value auctions have long been in existence and fiercely debated in the economic literature. The implications of this theory are far reaching for the fast emerging electronic auction markets. Businesses can make strategic uses of the auction markets ranging from estimation of demand curve for their products to achieving individual price discrimination and liquidating their inventory at the highest possible price. However, the implications of the possible existence of winner’s curse are far reaching for both businesses and consumers. In this paper, we seek to investigate the existence of winner’s curse and establish its determinants.

Financial markets data present a challenging opportunity for the learning of complex patterns not otherwise discernable, and machine learning techniques like genetic algorithms have been noted to be advantageous in this regard. Independent trials of the genetic algorithm are known to explore different parts of the search space and produce solutions which potentially capture different patterns in the data. Additionally, learning in domains prone to noisy data can generate solutions which obtain performance gains by fitting to what essentially is noise in the data. The article investigates possible strategies for combining the rules obtained from independent GA trials with the objective of noise filtering or enhanced pattern detection for improving the overall learning accuracy.