Richard Holowczak

An experimental study on content-based image classification for satellite image databases

Current art uses metadata associated with satellite images to facilitate their retrieval from image repositories. Typical metadata are geographic location, time, and data type. Because the metadata do not indicate which regions within an image are obscured by clouds, retrieval with such metadata may produce an image within which the region of interest (ROI) for the user is not visible. We report a system that can automatically determine whether an ROI is visible in the image, and can incorporate this into the metadata for individual images to enhance searching capability. The goal is to annotate each image with metadata regarding a number of ROIs. An experiment with the system annotated 236 advanced very high resolution radiometer (AVHRR) images of the North Atlantic from a five-month viewing period with descriptors that expressed the visibility of an ROI centered on Long Island, NY. For ground truth, we used the classifications of three human subjects to determine visibility of the same region of interest, and labeled the ROI with the majority decision of the three subjects. Partial cloud cover made the human determination subjective, and resulted in disagreements among the subjects. Using randomly selected training subsets of the images, we found the two images whose regions were most like those in images for which the Long Island region was visible.

Customized geospatial workflows for e-government services

The past decade has experienced a phenomenal growth in the electronic delivery of business services. This has led to an elevation in the expectations of citizens for fast and efficient delivery of governmental services. Recently, workflow systems have gained importance as an effective= infrastructure for automating the business processes within and across government agencies. Government services, such as permit processing for the development or preservation of land, can be modeled as workflow. They must consider geographic locations and the geodata that include location-specific data, documents and information. We call such workflows dependent upon locations and their geodata Geospatial workflows. We argue that geospatial workflows vary widely from one case to another, and therefore cannot be represented as a generic workflow. We demonstrate, with concrete examples, that there is a compelling need for customized geospatial workflows and propose a system that is capable of generating such customized workflows by capturing the user requirements and extracting the relevant governmental regulations. Our system also provides users with decision support functionalities comprising of various GIS layers, by which users can evaluate and identify a suitable workflow among various possible scenarios. Geospatial workflows can be used by government agencies to automate the delivery of their services. We present the features of a preliminary prototype.

Aggregating Information in Option Transactions

Implied volatility is a prime example of valuable information about an underlying stock that can be extracted from the market prices of its options. But the volatilitysmile illustrates the problem that there are many calls and puts, with different strikes and maturities, that can provide different information. How best to aggregate it is an important issue, especially for market makers. Some popular option-based measures, such as put-call ratios, simply add up numbers of contracts, with no adjustment for their different exposures to the factor one wishes to measure.In this article, the authors propose that investors classifyoption trades as buyer-initiated or seller-initiated, and useGreek letter exposures to combine them into an “aggregate delta order imbalance” (ADOI) to measure relative demand for upside and downside exposure to the underlying, and an “aggregate vega order imbalance” (AVOI) to measure demand for volatility exposure.Applying the approach with QQQQ options on the NASDAQ 100 index, tracking stock at intraday intervals, they show that both measures contain predictive information for the stock return and its volatility in the same period and in the periods immediately following.

Incorporating Real-Time Financial Data into Business Curricula.

The need to incorporate business and economic data into curricula has been a driver of technology adoption in business schools. Webbased resources and professional data services, such as Reuters and Bloomberg, are being increasingly adopted by business programs to meet this need. There are clear trade-offs to adopting either technological approach. In this article, the author presents examples of incorporating real-time data from professional data services into a variety of business topics.

The impact of media on how positive, negative, and neutral communicated affect influence unilateral concessions during negotiations

In the online environment, audio and instant messaging (IM) media are quite commonly used by people to communicate with each other and make offers as they negotiate. While we know much about how IM and audio differ, we know very little about how offers that are favorable to the recipient (termed unilateral concessions) are affected by what and how people communicate over these media. The purpose of this study is twofold – (1) to examine how such concessions are influenced by communication that is either neutral, or positive, or negative in affect; and (2) to determine how the use of IM, relative to the use of audio, influences the effects of these types of communication on unilateral concessions. We develop a research model based on prosocial theory, which suggests that negotiators using audio are predisposed to interpret their partners’ motivations and behaviors in a positive (prosocial) light while negotiators using IM are predisposed to interpret their partners’ motivations and behaviors in a negative (competitive) light. We manipulate the use of IM and audio in anexperiment designed to test predictions based on this theory. Our work provides theoretical and empirical support for the idea that communications other than concessions (such as positive, neutral, and negative affect) can lead to more or less self-sacrifice depending on the medium employed, and thereby motivate negotiators to make greater or fewer unilateral concessions. Specifically, we found that (1) positive affect comments can increase unilateral concession independent of the medium used by negotiators; (2) neutral affect comments can increase unilateral concession when negotiators use audio, but have little impact when they use IM; and (3) negative affect comments can decrease unilateral concession when negotiators use audio, but can increase unilateral concession when they use IM. These results provide insights to researchers and practical guidance for negotiators.

Data warehousing in environmental digital libraries

The purpose of this article is to present what we view as a new application of data warehousing in non-traditional domains such as environmental monitoring and research, biomedical applications, and other types of digital libraries (DL). While the work we present here focuses on an environmental data warehouse, the ideas are equally applicable to other DL environments. Data warehousing has been embraced by the professional IT community with many successful implementations reported in the retail, insurance, finance, and other industries [7]. For the most part, traditional data warehouses share the following common features. First, virtually all the data propagated from the operational systems and data sources to the data warehouse is either numeric or character-based. The data warehouse database contains mostly aggregated and summarized data in numeric format. Next, operational systems and data sources are identified early on in the warehouse development cycle. Once the warehouse is operational, new data sources are often difficult to integrate into the data warehouse schema. Quite often the warehouse database is designed around a core set of business needs that are decided upon in advance. The set of queries that can be addressed by the warehouse might therefore be fixed. By contrast, data warehouses to support DLs must contend with a variety of multimedia data types from a collection of component systems that may grow over time.

Using Reuters 3000 Xtra system for financial information education

Reuters 3000 Xtra is a real‐time and interactive global news and financial information service that covers the equity market, bond market, foreign currency and money market, and various derivative markets. While there are hundreds of business schools in the USA, only a handful of schools have established real‐time financial and trading services for educational purposes. Currently only six schools are equipped with such services in the USA. The Subotnick Financial Services Center at the Zicklin School of Business of the City University of New York is the only real‐time trading facility for education in New York City. Given the many advantages of educating and training students using simulation of real‐time trading, the number of business schools that are considering acquiring such services is bound to increase. The emerging issue is how to use services such as Reuters 3000 Xtra to provide financial information education for students, faculty members, and potential traders. This article covers some important features of Reuters 3000 Xtra and illustrates how it may be used to provide financial information education. This article specifically discusses the use of Reuters Instruction Code (RIC), Speed Guide, Model Browser, and PowerPlus Pro to access, retrieve, organise, display, and analyse real‐time data. Furthermore, this article demonstrates taking advantage of the real‐time data feed to test the portfolio diversification theory by developing a diversified portfolio in the equity market. It is our hope that students, faculty members, business librarians, business information specialists and financial portfolio managers who are involved in Reuters 3000 Xtra training and teaching will benefit from this article.

Concept based query of digital library objects

Research in digital libraries has many applications in the field of biomedical information. Our current research into the areas of concept extraction, indexing and searching in digital libraries has led to novel applications in information retrieval and indexing. We present an overview of our work and discuss future applications in biomedicine.

Consolidating Information in Option Transactions

Underlying each stock trades hundreds of options at different strike prices and maturities. The order flows from these option transactions reveal important information about the underlying stock price. How to aggregate the trade information of different option contracts underlying the same stock presents an interesting and important question for developing microstructure theories and price discovery mechanisms in the derivatives markets. This paper takes options on QQQQ, the Nasdaq 100 tracking stock, as an example and examines different order flow consolidation mechanisms in terms of their effectiveness in extracting information about the underlying stock price and volatility movements. The analysis leads us to propose an aggregation weighting scheme that depends both on the liquidity of each option contract and the contracts risk exposure, delta for stock price movement information and vega for volatility movement information. Based on this weighting scheme, we identify significantly positive correlations between the aggregate option order flows and the realized returns and volatilities. In particular, the delta buy pressure positively predicts the underlying return and the vega buy pressure positively predicts the change of volatilities.

Price Discovery and Liquidity Characteristics for U.S. Electronic Futures and ETF Markets

Using high-frequency datasets, we examine price discovery and its determinants for equivalent instruments across futures markets, electronically traded exchange-traded funds (ETFs), and spot markets. We compare futures to ETFs—leveraged and unleveraged—for stock indexes, using both a normal period and the 2008 financial crisis. Yan and Zivot’s information leadership procedure is employed to determine which instrument dominates price discovery. We then examine the determinants and characteristics of the price discovery process using Hasbrouck’s sequential trading model for the price impact of large trades. We find that most price discovery occurs in the more liquid and highly leveraged futures market. Although liquidity declined in all markets during the financial crisis, the relative contribution of ETFs to price discovery increased. We also find that the information leadership shares of futures and ETFs depend on the ratio of the quoted percentage spread between futures and ETFs and the aggregate volatility occurring in these markets.