• General information
• Educational opportunities
• Projects and activities
• Jobs and career information
• Publications
• Faculty and staff
• Alumni Connection
• Archives

Data Warehousing and Intelligent Agents: Tools for the Future

Dr. Mehdi Owrang

October 7, 1998

John Studach

Database Management

64.635

I Introduction - (This paper outlines an approach that is part of a continuum of strategies used to influence behavior).

Visionaries of the Information Age from Vannevar Bush (Bush, 1945) to Nicholas Negroponte (Negroponte , 1996) have inspired people to look critically at the impact of technology on their lives. Businesses, academics, and the general public have been intrigued by the plethora of new and exciting opportunities that are now possible because of the recent advancements in technology.

In the Powershift: Knowledge, Wealth and Violence at the Edge of the 21st Century, Toffler stated, "the neglected fact that big breakthroughs often come not from a single isolated technology but from imaginative juxtapositions or combinations of several of them" (Toffler, 1990). A phenomenal number of practical applications have emerged because of the confluence of the computing and telecommunication industry. Often the new market space that is created by the intersection of two or more unrelated or previously unexploited technologies results in powerful and exciting business opportunities.

Recently, a new breed of applications based on database technology and sophisticated datamining techniques have been developed to exploit the ubiquitous nature of the Internet. Sophisticated artificial intelligence (AI) tools greatly enhance the range, scope, and quality of the data that is available.

Data is collected and managed through technology. It is then transformed into information. The resulting knowledge becomes power. Rob and Coronel state that, "data mining technologies have the potential of becoming the next frontier in database development" (Rob and Coronel, 1997, p. 733).

The Internet has become a juggernaut; a powerful new access and delivery medium. The Internet has become the embodiment of McLuhan's mantra, "the medium is the message" (McLuhan, 1967). The "Net" has opened up new dimensions for developers, businesses, practitioners, and the general public. Entrepreneurs are anxious to exploit the enabling power of the technology. However, the relentless pace of development in technology often means that developers are searching for the most effective and successful models at the same time the systems are being deployed in the marketplace. One fact that has become particularly evident is that models that have worked well in one domain are not necessarily effective when two or more technologies are combined.

Many developers are desperately searching for the next "killer app". Entrepreneurs are focused on finding ways to use, exploit, or develop business opportunities. Companies are seeking ways to collaborate with others or to develop their own niche in the marketplace. Some times a powerful synergistic effected can be created when the right two technology components are connected into a system that is linked via a global network. Several examples of these systems will be discussed in this paper.

Three fundamental principles of the Information Age will be featured in the discussion that follows. First, data can be converted to knowledge. Next, Information is power. Third, technology is an enabler. Sophisticated systems can be developed to capitalize on these three elements. Cutting-edge systems that incorporate each of the telecommunications, computing, and information systems domains are being deployed. Some of the most sophisticated ones are employing intelligent agents, meta-databases, and datamining techniques. Some individuals are forecasting that this new paradigm has revolutionized much of the industry and the way business is done.

The purpose of this paper is to provide a summary of the current state-of-affairs in the field of market segmentation, precision customization, and datamining techniques over the Internet. Basic technology components, principles, business assumptions, implementation, and delivery models that apply to databases will be discussed. Many of the relevant issues will be identified and the implication of several of them will be explored in depth. Finally, the developments in this field will be explored for on-line health promoting systems of the future.

Three prominent companies that have been recognized for their pioneering efforts will be featured in this paper. The systems will be used to illustrate the practical application of the concepts, as well as provide a context for evaluating their effectiveness. Some of the areas for downsides and areas for future development and research will be discussed.

The sections that follow will provide some of the answers to the basic who, what, where, when, why, and how questions. The second section will highlight some of the ways that companies are exploiting database technology through networks. The theories, models, concepts, and principles will be covered in the third section. The technology components will be discussed in the fourth section. Conclusions, implications, and recommendations for future directions will be the final part of the paper.

II Customization, Personalization, and Tailoring

A. The Business Case

The success of failure of many business ventures depends on quality of the relationship that exists between the customer and the seller (Pyle, 1998). Because of the intense competition in today's global marketplace, the philosophy of Henry Ford (they can own a Model T in any color as long as it's black Weiss, 1988 p. 67) is way no longer a viable option. Today, "the customer is always right" has become the motto for many companies. Often the entire business strategy for companies is based on knowing whom their current and potential customers are, what their needs, attitudes, perceptions, and buying patterns are, and how they change over time. Their primary business goal is to develop a close long-term relationship with the customer. It is the job of the business to customize their products and services in response to the desires and changing buying patterns over time, not the responsibility of the customer to conform to the company. For companies who have diverse product lines and who are dealing in the global marketplace, capturing data about their customers is an enormous challenge.

Some companies such as Safeway, Dell, Lycos, and Amazon.com have been more successful at closing the gap between the customer and vendor. By developing data warehouses and employing datamining techniques, enterprises have a much better idea of their customer base and their buying patterns. They are in a much better position to develop "one-to-one relationships, a million at a time" (Micromass, 1997).

Business who are conducting on-line often use a methodology that calls for compiling a personal profile through a front-end interface, and varying the content at back-end. Some of the more sophisticated tactics and approaches use dynamic and intelligent agent techniques that can match preferences with use or buying patterns with great precision (Weiss, 1994). With an eye to the future, several start-up companies are using algorithms to predict the buying patterns from data that reside in multiple third-party or megaadata warehouses.

B. The steps

The process of customization is similar for most organizations. Among the first steps is to establish, align, or articulate a strategy that makes identification and tracking of customer preferences and habits a core element of the business process (Stair, 1996). The goal is to personify a customer-centric attitude. Tactical decisions are based on this approach.

Ideally, the sales, marketing, and customer service departments work closely with their counterparts in information systems to develop systems that are effective, efficient, manageable, and cost effective (Berson, 1996). Point-of-sale or point-of-contact data are captured in a way that does not aggravate their customers or feels minimally intrusive. For example, Safeway and Amazon.com have systems that capture the buying patterns of their customers seamlessly.

In order to broaden and strengthen the customer base additional data is gathered through a variety of schemes and third-party sources. Various promotions, special events, and marketing campaigns provide supplemental information on current as well as identify potential new customers.

External data vendors are also a rich source of data. Several national data warehouses capture information from census tracks, demographic profiles, as well as public records and private vendors. These companies manage and analyze the data to develop profiles about targeted market segments, "clusters", or specific customer. Weiss estimates that there are over 55,000 subscriber lists as well as census track data, and a host market research databases that are being joined to extend coverage and a more thorough understanding of the customer base (Weiss, 1988).

Once the customer profiles are developed, companies can develop new products, revise existing ones, and develop or adjust marketing strategies to match products to buyer characteristics or habits (Claritas, 1998). The success of these efforts is often a function of quantity and quality of the data that is collected.

Many innovative techniques have been used to gather data. Special promotions and registrations are used to populate databases. Data is frequently culled from information that is part of the transaction or method of payment (Rob and Coronel, 1997). First-time on-line data is collected and repeat users are identified through "cookies" (Donaldson and Seigel, 1997). More organizations are now storing the data in databases that are updated dynamically after each transaction or site visit.

III Theory, Concepts and Principles

The fundamental principles of market segmentation or precision targeting are drawn from several fields of study. The three most important domains are business, information systems, and the psychosocial arena. The integration and application of the basic principles from each field must be considered during the initial stages of the planning process.

The customization process is heavily dependent on information systems concepts and application of sophisticated database technologies. The overall process includes aspects from all three fields and the end product is built on the precepts of each. The principles of business and psychology can be conceptualized as a form of "middleware" through which the data and information is transformed and translated into practical applications.

Many subdomains are subsumed within IS. However, an in-depth discussion and analysis of each is beyond the scope of this paper. The discussion of the various components will be very general since each of the systems described below employ unique approaches to accomplish the business objectives.

Data warehousing and data mining concepts are the primary elements of IS that relate to customization. The following are some of the questions that relate to the building blocks of the data wareshouse systems:

• Database design, management, administration, and evaluation ­ What types of databases are most appropriate for collection, storage, and accessing the data? Are relational, object-oriented databases or data marts appropriate? What design and structure issues will allow for optimal utility of the database? What are the relevant flexibility, standardization, extensibility, and interoperability issues? What are the minimal performance issues such as frequency of access, speed of queries, number of users, and access patterns? How important is concurrency? What are the backup and disaster data recovery plans? What are the implications for privacy and security of the data? What skill sets do the IS employees have?
• Database tools and techniques such as datamining and data warehouses ­ How can data integrity be maximized? What are the acceptable levels of data integrity? What are the needs for migration of data from other legacy databases or external sources? What management tools are most appropriate and how comprehensive or sophisticated should they be?
• Database software ­ What type(s) of software is most appropriate for the project? Is it proprietary or open standards and compatible with other systems along the data provider chain? What are the management and administration needs? What are the restrictions and advantages of each for distributed databases and multiparty data repositories? What are the training and support needs of the staff who will implement and monitor the software?
• Networking and telecommunications ­ How much data needs to be moved, how often, and when? How critical is speed? How large is the organization? What is the potential for expansion? What operating systems, and network operating systems are uses? What is the potential for data gridlock and bottlenecks? What are the weak points in the system?
• Systems design, infrastructure, and architecture ­ What are the configuration and design issues? What is currently in place? What are the projected growth needs? What is the nature of the client server architecture? How do legacy systems impact on the business goals? How much of the data is accessed in real-time? How much On-Line Analytical Processing (OLAP) takes place?
• Artificial Intelligence, Expert Systems, and Neural Networks ­ How can the more sophisticated techniques be employed to enhance the systems? What resources and employees with the appropriate skills are available for the design, implementation, and maintenance of the advanced tools? Where are the concentrations heaviest use? What are the prospects for the future?
• Human factors ­ What are the "human" issues related to designing, implementing, and administering the system? How are work and design groups formed? What are the issues related to group and corporate structure? How are training and support handled? What are the levels of empowerment? How can user interfaces be enhanced?
• Evaluation and Research ­ What kind of data needs to be collected to determine the efficacy of the systems? How much data needs to be collected? Who will compile the reports? How important are the system performance metrics to management or the organization?

Although each of the cases cited below employ different approaches and tactics, each of them will follow a process. Multiple-processing-multiple-site (MPMD) databases have been used in each of the cases because of the complex nature of the vendor relationships and the comprehensiveness of the approaches. Today, most very large databases (VLDB) use relational or object-oriented models.

The first step in the process is to define the business goals and objectives. A careful analysis of the business needs and resources is conducted. Dedicated databases are designed or created to support the objectives by collecting, storing, and managing new or old data. The stragetic analysis takes into account people, resources, technical, and management issues for the short as well as the complete lifecycle of the database.

The second step is developing a model that is appropriate for the business problem. Entity-relationship modeling tools are used to represent the database and the data, variable, values, attributes, fields, and relationships it contains. Techniques such as normalization, indexing, bitmaps, partitioning, replication, and aggregation are built into the design of the system to ensure data integrity and optimal levels of performance.

The database is populated with primary or secondary data. Primary data is collected directly through the purchasing patterns of the customer base. Surveys, feedback, and interviews provide additional information about customer preferences, perceptions, thought processes, and decision making. Secondary data is collected through third-party sources such as census data, market profile agencies, or other entities along the vendor chain. A new wave of on-line tools such as cookies, robots, spiders, and agents add very important individual and aggregate data (Claritas, 1998).

Analysis is the next logical step. The data is explored with ad hoc queries or though automatic data mining techniques. These transactions take place in real-time, often using the OLAP methodology.

When the data is stored in data warehouse, data mining techniques can be implemented. In Chapter 13 of Database Systems, Rob and Coronel describe data mining as a proactive approach. They define four phases of data mining; preparation, analysis and classification, knowledge acquisition, and prognosis. Data mining tools automatically search for the data for relationships, or unexpected problems. The data mining tools can uncover hidden opportunities in the relationships among the data. Models and processes that require minimal end user intervention are designed to predict behavior. The data mining tools are based on algorithms. The algorithms serve as "the building blocks for AI, neural networks, inductive rules, and predictive logic" (Rob and Coronel, p. 730). The data-mining tools applies specific algorithms to find data grouping, classifications, clusters, or sequences, data dependencies, links, or relationships data patterns, trends, and deviations. When the full extent of organizational data is explored, powerful applications of corporate intelligence are possible. The next generation of intelligence can be extracted from the data.

Many of the AI techniques of the customization process are based on the recognition of patterns in the data. The data may be numeric, alphanumeric, or text. Pattern recognition and semantic usage in natural languages is based on two prominent theories. Bayes theorem calculates the probabilistic relationship between multiple variables. In Shannon's Information Theory, information can be treated as quantifiable variable. It states that the less frequently a unit of communication occurs, the more information it conveys. Therefore, the amount of usage establishes relevancy.

A number of negative attributes to data mining have been cited in the literature (Claritas, 1998; Neilsen, 1998). The complexity and comprehensiveness of the systems have inherent issues. "Big Brotherism" is another critical issue that is often mentioned by privacy advocates. Security of data is another closely related problem. As databases are constructed and segmented difficulties in connecting the "islands of information" arise (Rob and Coronel, 1997).

The relevant business concepts include the organizational structure, process development, deployment, and marketing aspects. An in-depth understanding of the customer, their perceptions, and decision-making process is achieved by analyzing key geodemographic and psychogenic variables (Weiss, 1994). The ultimate goal of the marketing strategy is to strive for a one-to-one relationship with the customer (Peppers and Rogers, 1998).

Theories from psychology and sociology help systems designers and marketers understand the mindset of customer base. Careful market analysis can provide insight into the dynamics and decision making process at the individual, group, cultural, or national level. That knowledge is critical in knowing how to approach and relate to the customer base throughout the length of the buyer-vendor relationship. Social marketing incorporates marketing principles from business and social theory to influence behavior in non-market driven campaigns (Andreasen, 1995).

IV Database and Datamining Technology

Three companies have been selected to illustrate the application of leading-edge database and datamining technologies. Each of the companies uses these sophisticated techniques and applications in innovative ways. The ultimate goal of each of them is to get close to the customer. Firefly Networks use a fully aggregated database complex and neural networks to develop the collaborative filtering process (Firefly, 1998). Claritas uses datamining as a process to form a precision targeting methodology which it provides to a variety of customers across the spectrum of marketing agencies (Claritas, 1998). Autonomy Inc. uses concept matching of unstructured digital information as a means to provide dynamic personalization of information (Autonomy, 1998). For each company, the process starts with the collection of data and then using powerful tools to turn the data into usable knowledge for its company.

A. Firefly Network ­ Firefly.com

The Firefly Network emerged from research that was conducted at Massachusetts Institute of Technology's Media Lab. Firefly began as Agents, Inc. and was incorporated in March, 1995. A 1996 press release stated that

Agents, Inc. was working with strategic partners to deliver "breakthrough" technology and "community" to the Web (Firefly, 1996). Based on the forecast that "1996 will be the year of the intelligent agent", Firefly went live on the Web in October 1995 (Firefly, 1996).

Firefly.com has achieved significant recognition as a leader for providing flexible agent technology for the on-line sector of the economy. The Firefly Network Inc. personalization software captures and adds preference and general interest functionality to businesses that use its technology. Firefly has the only US patent on "collaborative filtering" (Dragan, 1997). The rules-based processing is used to match content according to predefined rules. Neural network algorithms are invoked based on user actions. A unique feature of Firefly is their fundamental approach is to provide content that is customized from user profiles instead of the producers preconceived notions.

The most powerful part of artificial intelligence is the neural networks' recommendation systems. The systems are based on recommendation engines that take actions and gets profiles based on explicit actions to implicit rules. The product relies on CGI scripts and libraries working with any ODBC-enabled database (Dragan, 1997). This open systems approach facilitates multi-vendor compability.

The Firefly process is called "collaborative filtering" (Mattis and Ubois, 1998). Firefly users enter the system either by registering at Firefly.com for a free "Passport", or by visiting one of the affiliated sites where they fill out a forms-based questionnaire. Firefly uses a 1-7 rating scheme. Clients are asked to rate items from the Website such as movies, articles, or books. When they have rated at least six items a database collaborative community profile is established. The data is collected in a database and potential relationships are suggested or matched for like-minded individuals among the on-line community. Over time, the recommendation system will increase in accuracy as databases share data across multiple interest groups (Vonder Haar, 1997).

The recommendation system technology is called Advanced Collaborative Filtering (ACF). Webmasters can implement ACF by using HTML extensions of Firefly's object-oriented C++ API's (Dragan, 1997). The engine tracks the time consumers spend on each page and offers suggestions from related sites on the basis of the previously developed personal profile. Collaborative filtering is much faster and efficient than artificial intelligence approaches that use neural networks and rules-based processing. It actually mirrors the critical thinking process (Dragan, 1997).

Firefly markets itself as a system that is able to reach customers for one-tenth of the cost of traditional advertising. For economic reasons, that is one of the major reasons why this technology is so appealing to on-line vendors. Much of the guesswork is taken out of the "shotgun" marketing and sales operations when companies know the specific preferences of its potential consumer base. By enabling Web publishers to create larger databases across multiple vendor sites, further customization is possible. Being part of the Firefly network, enables companies to integrate the Firefly's system into their software and add value to their site.

Several issues have been raised about the Firefly technology (Neilsen, 1998; Firefly, 1998). Privacy, security, and proprietary systems are three issues that are most frequently named. Firefly has advocated for an open standards approach since its inception. In 1997, Firefly submitted an Open Profiling Standard (OPS) to the WWW consortium for approval to allay the concerns of privacy advocates (Dragan, 1998).

Firefly has licensing, affiliation, and cooperative agreements with several other business groups. Gustos Software LLC uses a recommendation engine for collaborative filtering for users to rate Web sites.

The LikeMinds Inc. Corp. uses collaborative filtering and predictive modeling technology to tap into customer behavior data. The LikeMinds technology agents monitor the on-line behavior of users and import the data into databases. That data is used to matched product selections that have the "best chance" of being purchased.

A third company, Net Perceptions Inc. uses recommendation engines.

Imana Inc. agent technology offers personalization information delivery and targeted advertising and self-forming on-line communities (Mattis and Ubois, 1998).

Alexa Internet takes information from third-party providers to personalize offerings to customers. It uses a "surf engine" to monitor the access patterns of individuals and make recommendations for similar sites. (Williams, 1998)

In 1998, Lycos paid $39 million to acquire Wise Wire Corporation (Blanchard, PC/AI, 1998). It professes to use smart filtering to provide real-time content presnetation based on organization objectives. Moreover, it attempts to build "community" around shared interests. Open standards-based platforms for personalization intelligent agent technology are the key components of the system.

Each of the companies above uses a slightly different database architecture to accomplish its objectives. The configuration of the database of three of the groups that are affiliated with Firefly are illustrated in Appendix I.

B. Claritas ­ claritas.com

Claritas has been in business since the 1970's. It has established itself as the industry leader in market segmentation information. Claritas sets the standard for integrated marketing. Much of Claritas' work is based on the work of its' founder Jonathan Robbins who declared, "Tell me someone's zip code, and I can predict what they eat, drink, drive ­ even think" (Weiss, 1988).

Robbins developed the (Potential Rating Index for Zip Marketers) PRIZM database application to predict individual preferences and buying patterns based on the where they lived. His premise is that "birds of a feather flock together". Like-minded people tend to live in similar communities and their tastes and consumer preferences are remarkably alike and stable for ten-year periods (Weiss, 1998).

PRIZM uses algorithms to match individuals by zip code with their consumption patterns and lifestyle habits. Claritas developed a matrix of 40 clusters based on several sociodemographic variables. The two primary axis of the matrix are population density (urban ­ rural) and personal wealth. Initially, Robbins used census tract data to develop the 40 clusters. He used pubic and third-party databases such as TRW and Infobase to determine consumption patterns and preferences. His algorithms were derived from the correlational relationships between the zip codes and the lifestyle variables. There are regular and periodic updates of the core databases. Marketing techniques such as surveys, interviews, and focus groups provide additional information are used to enhance the predictive power of the algorithms.

Claritas sells the aggregated data in reports, as well as software that represents the profiles for target markets by locale, lifestyle, market potential. Claritas states that the precision targeting and lifestyle segmentation is accurate at the micro-neighborhood level. The lifestyle portraits contain integrated product, media, and lifestyle information. Corporations, small businesses, and direct mail marketing agencies use the data (Claritas, 1998b).

Claritas offer segmentation data on diverse populations such as Hispanic Portraits and Workplace PRIZM which targets the workforce that populates a city only during the day. All of the traditional media channels are integrated into the database and are linked with agencies such as Arbitron and the Neilsen rating systems. Marketers use the data to develop market potential as well as customer loyalty (Dwek, 1998)

Claritas offers its services in Canada, and has recently opened offices in England and Europe. It uses its comprehensive Website to market its products.

C. Autonomy - agentware.com

The products developed by Autonomy are based on the of concept matching of unstructured digital information with a goal of providing dynamic personalization of information. It evolved from the neural network and pattern-matching technologies developed at Cambridge University (Autonomy, 1998). Autonomy has developed software for the intelligent management of on-line content. It uses agents to gather data and promote "knowledge transfer" within organizations. By tracking on-line accessing patterns of individuals through cookies and matching it with psychographics data, very accurate profiles can be developed for each user. The concept matching engine queries the database and then presents personalized content in real-time (Autonomy, 1998).

Automatic agent technology is used to analyze the content of WebPages and categorize the unstructured information into user defined content and concept channels. The agents "autosuggest" similar links to individuals or others who have similar interest. Matching is accomplished for the individual as well as like-minded users throughout the identified on-line community. A set of interest profiles based on the most common and important subjects sought by the visitors at a site is developed and becomes incorporated into the concept matching engine. The ultimate goal of this process is to foster user loyalty at the site.

A knowledge server is used to connect people, knowledge and insight. The query engine incorporates techniques such as natural language, Boolean, Proximity, Agent, Wildcard, Fuzzy and concept querying (Autonomy, 1997) A Dynamic Reason Engine (DRE) uses the neural network to look for similar patterns, which develop key concepts/ideas. The DRE tracks shifts in content and matches users based on predefined categories. The query is language independent and multiple languages can be accommodated in a single engine. The data in the database can be query by date, which allows for very sophisticated presentation schemes. All of the API's have been modularized into function suites. The Indexer provides automatic extraction of key concepts from documents and allows for high speed indexing. Several features such as "autoelimination" of duplication, and "autosummarization" have been integrated to promote high performance.

The open index file format, configurable index options, and multiple concept indexes per engine provide a great deal of flexibility in the systems. The Adaptive Probabilistic Concept Modeling (APCM) is based on the Bayes and Shannon's Information Theories. The server takes the input and determines where new documents will fit best. This is called "precision tagging". Meta tags are used for import/index processes.

Autonomy is very concerned about user authentication and security. It also supports for proxy servers and includes firewall protection.

Autonomy functions on multiple platforms for content servers. The open-systems archiceture allow the company to leverage its' existing infrastructure (Autonomy, 1998b). Recommended system requirements are:

Solaris/SPARC 2.5

128 Mb of RAM

1GB hard disk

Web Server

WindowsNT/Pentium

200 Mhz Pentium processor

64 Mb RAM

1 Gb hard disk

Web server

DEC UNIX/Alpha 4.0

IRIX (SGI)

Any POSIX compliant version of UNIX

D. Health companies

The use of datamining and AI techniques in the healthcare sector is in the infancy stage. The healthcare debates of 1992 put enormous pressure on the health and medical communities to change the way it does business. The attention from the debates caused many businesses to look at the healthcare arena as a viable area for expansion. Very few examples are available where these sophisticated systems have been deployed (Studach, 1998).

Several reasons are cited in the literature. The healthcare sector is heavily laden with late adopters (Rogers, 1995). Another factor is that the mindset of the professionals in the field has been to try to do more with less people. Often adding IS personnel to solve business problems is not considered. Most of the development has occurred in administrative and clerical tasks, instead of innovative approaches. A third factor worth noting is the caviler attitude of many hospitals, provider organizations, and governmental agencies (Ferguson, 1997). A customer-centric approach is a rare business practice.

Other business reasons contribute to the complacency in the industry. Some of the hospitals and HMO's are so mired in reengineering their work processes, they have not had time to investigate datamining and customization. Many of the affiliated providers are in the non-profit arena. They do not have the financial resources or clout to attract or maintain database technologies of the past. However, with the advent of the new generation of tools, some of the foreword thinking organizations have begun to move in the direction of datamining and tailored interventions.

The American Heart Association has recently developed a prototype for a database system that will allow for tailoring health information on the basis of health risk, personal preference, and demographic data. The content that is presneted to the user will varied on the basis of expert systems that rely on proprietary algorithms (Bazzarre, 1998). The database design and algorithms are provided by Micromass.com.

Globalmedic is a Canadian company that is using expert systems and multiple object-oriented vendor databases to provide customized content to subscribers. The company was started in 1993 with $100 million of venture capital. They provide individual behavioral change information as well as composite approaches for companies through their intranets (GlobalMedic, 1998). The databases reside on Microsoft Windows and NT platforms and use Microsoft SQL 6.5.

The University of Michigans' Comprehensive Cancer Center has been a strong advocate for the application of tailoring in its interventions. Until recently the majority of its efforts have been devoted to operationalizing interactive kiosks that employ tailoring techniques. An on-line version has recently been developed. Plans for the future include the on-line collection of data that will become part of a enterprise-wide data warehouse (Strecher, 1997).

Intellihealth has teamed with PointCast in an attempt to present customized content to users. The users provide information about their information preferences, and agents supply the appropriate content. No information is currently avaiable about plans for data warehouses (Partnerships, 1997).

V Conclusions, Implications, and Future Directions

A. Implications

When health and business practitioners learn and become aware of how some of the new technologies are being applied in various sectors of the economy they can frequently get a glimpse of how technology will be exploited in the future. However, this paper has concentrated more on what is possible that what is probable in the foreseeable future. This arena is very much in its infancy. Many of the basic terms have not been standardized. For example, A well respected authority on Web design, Jakob Neilsen uses customization as a term to describe computer preferences that are adjusted by the user, and personalization as the adjustment of the preferences by the computer (Neilsen, 1998).

Further, Neilsen contends that computers can never match an individuals ability to pick the topics in which they are most interested. In a simple world where the amount of information is small, that statement will always be the case. However, most individuals who are active participants on-line, are often stunned by the plethora of information that is available to them. Most users very quickly become aware of the meaning of information overload.

Businesses, marketeers, and content providers are becoming increasingly aware of the relative ineffectiveness of "shotgun" approaches. Those approaches were acceptable in the days when any content had to be manipulated manually. Anyone who works or is living in the information age, knows that the enabling power of technology can be used to do what was not possible with "manpower" only. Marketers use precision targeting to gain market share. Flexible media channels are using "narrowcasting" rather than broadcasting.

Technology is is advancing rapidly. New and and innovative tools are getting more sophisticated and easier to use every day. The concept of personalizing every realtionship is a powerful idea. Every bit of data can be viewed as a corporate asset. When the data is analyzed over time the strenght of the predictive power of historical data increases.

The advancements in the telecommunications industry has opened up new delivery channels and medium. Customization and the application of technologies based on data warehousing are possible for every new innovation, or at the confluence of two or more technology developments. In the new global order, we should learn to view the world as one mega-enterprise. The VLDB object-oriented databases can be mined to to provide multimodal customizations and presentations.

This new order of business is not without threats, concerns, or downsides. The invasion of privacy is at the top of most people's list. The threat of "Big Brotherism" is a serious concern for many individuals. They do not want their personal data and profiles collected or being passed electronically withouth their knowledge or consent. Most consumer groups agree that consumer should be in control of their data. In the near future, policy discussions will focus on the development of a viable code or on-line bill of rights. However, as always, the adage, "let the user beware" needs to the approach for consumers. This is particularly important in the medical and health area where there is a great deal of sensistivity about ones health records.

Technology has the potential to change our lives and the way we do business the same way that the clock, railroad, electricity, cars, and airplanes changed the lives of previous generations. The merger of telecommunications, database technology, and AI opens up many new and exciting horizons. As IS professionals, we must apply the techology with a great deal of care and wisdom.

References

1. Andreasen, A. R. (1995). Marketing Social Change : Changing Behavior to Promote Health, Social Development, and the Environment. San Francisco, California: Jossey-Bass Publications.

2. Autonomy, (1998) http://www.agentware.com.

3. Berson, A. (1996). Client/Server Archetecture. . (2th Edition),New York: McGraw-Hill.

4. Blanchard, D., (July/August, 1998) Lycos Gets WiseWire, PC/AI, 33.

5. Bush, V. (1945, July). As We May Think. Atlantic, 101-108.

6. Claritas. PRIZM. 1998, September 13, http://www.claritas.com/prizm.htm.

7. Department of Health and Human Services. (1997, May). Partnerships for Networked Consumer Health Information Conference, 1997, Washington, DC, 1997, April 14 Washington, DC: Department of Health and Human Services.

8. Donaldson, S. E. Siegel, S. G. (1997) Cultivating successful software development : a practitioner's view: Upper Saddle River, N.J. : Prentice Hall.

9. Dragan,R. V. (199) Firefly Network Inc. Passport Office Recommendation Systems http://www.agentware.com.

10. Ferguson, T. Workshop on Interactive Technology and self-care. 1997, April 17,

11. Firefly. Firefly. 1998, September 8, http://www.firefly.com.

12. Mattis, M., Ubois J. (January, 29 1998). Recommendation Systems/Collaborative Filtering. Upside Today http://www.upside.com.

13. McLuhan, M. (1967). The Medium Is the Massage. New York, New York: Bantam Books.

14. Micromass Communications, I. IntelliWeb Primer: Web Tailoring for Personalization Technology Overview. 1997, http://www.micromass.com.

15. Neilsen, J. Alterbox. 1998, October 4, http://www.useit.com.

16. Peppers, D., Rogers, M. et al. (1998). Enterprise One to One : Tools for Competing in the Interactive Age. New York, New York: Currency Doubleday.

17. Pyle, D., (September/October, 1998) Knowledge Discovery and Data Mining: The Expectation of Magic, PC/AI, 14.

18. Rob, P., Coronel, C. (1997). Database Systems; ;Design, Implementation, and Management. Cambridge, Masachusetts: International Thompson Publishing.

19. Rogers, E. M. (1995). Diffusion of Innovations. (4th Edition),New York: Free Press.

20. Stair, R. M. (1996). Principles of information systems : a managerial approach. 2nd ed. Danvers, Mass : Boyd & Fraser Publishers.

21. Strecher, V. J. (1997, September 30). Michigan Interactive Health Kiosk Demonstration Project: Annual Report October 1, 1996 - September 30, 1997. Ann Arbor, Michigan: University of Michigan Comprehensive Cancer Center Health Media Research Laboratory.

22. Studach, J. Health Promotion Informatics; A Peek into the Future. 1998, August 6,

23. Toffler, A. (1990). Powershift: Knowledge, Wealth and Violence at the Edge of the 21st Century New York, New York: Bantam Books.

24. Vonder Haar, S. (March, 3, 1997). Firefly Hopes Its Agent Plan Flies. Inter@ctive Week. http://www.zdnet.com/interweek.

25. Weiss, M. J. (1988). The Clustering of America. New York, New York: Harper and Row Publishers.

26. Weiss, M. J. (1994). Latitudes and Attitudes: An Atlas of American Tastes, Trends, Politics, and Passions. Boston, Massachusetts: Little, Brown, and Company.

Last updated on December 31, 1998.

You can send e-mail to Me.

Return to my papers or my dissertation page.