Rivero L.Encyclopedia of database technologies and applications.2006

Nikhilesh Dholakia

University of Rhode Island, USA

Detlev Zwick

York University, Canada

Anil Pandya

Northeastern Illinois University, USA

ELECTRONIC MARKETS

With continuing global growth of electronic commerce, consumers have come to exist in electronic “marketspaces” (Rayport & Sviokla, 1994) that are frequently surveilled through database techniques. This type of data-driven monitoring has been characterized as “dataveillance.”

This article reviews the main characteristics of such dataveilled and panoptic marketspaces, from the perspectives of critical social theories.

EXAMPLES OF DATAVEILLED MARKETSPACES

The term dataveillance is attributed to Australian computer scientist Roger Clarke (1988). Apart from the technically informed critiques by Clarke, the virtual dataspaces intersecting with human bodies have been discussed by Kroker and Weinstein (1994) and the condition of humans located in database matrices has been critiqued by Poster (1990a, 1995a). Aspects of dataveillance in e-commerce and m-commerce marketspaces have been addressed by Dholakia and Zwick (2001) and Zwick and Dholakia (2004).

What do the emergent dataveilled marketspaces look and feel like? We provide short descriptive cases to illustrate them.

JetBlue Passengers Face the Blues

The U.S. Department of Defense carried out a project to identify potential terrorists among airline passengers who could possibly attack army bases. Ordinary airline passengers became the unwitting pawns in this datainvasion exercise. In September, 2002, Torch Concepts— a data mining firm on contract from the army—acquired from JetBlue Airways (a discount air carrier) the itinerary

data for over 1.5 million passengers, including passenger names, addresses, and phone numbers. Of these passenger records, 40% were cross-referenced with gender, home specifics (owner/renter, etc.), years at residence, economic status (income, etc.), number of children, Social Security number, number of adults, occupation, and vehicle information—detailed demographic data obtained from a database vendor. While this substantial panoptic exercise—undertaken without any knowledge or consent of the passengers concerned and in violation of JetBlue Airways declared privacy policies—found only one possible anomalous passenger record, the potential for massive invasion of marketspace data records is quite evident in this case (EPIC, 2003).

Do the Insured Have a Clue?

Michael Ha of National Underwriter is worried that when massive databases such as the Department of Motor Vehicles (DMV) records and Comprehensive Loss Underwriting Exchange (CLUE) cross-reference each other, insurance companies can benefit at the expense of clueless drivers. Although such databases have legitimate uses, such as determining insurance premium rates for motorists, they can also be misused. Blending such private data with data collected from emerging technologies such as event recorders and global positioning systems, these systems are capable of generating real-time data profiles of drivers, in effect providing insurers a secretive “eye in the sky” to monitor drivers. When such data is married with information in massive DMV and CLUE databases, serious privacy concerns arise. Some insurance companies are encouraging automobile makers to install monitoring technologies in high-end vehicles. Ethical boundaries are being crossed, however. Allstate, a major insurer, has already paid $1 million to the California DMV in settlement of a dispute about unethical use of DMV records (Ha, 2003).

Dataveillance and Panoptic Marketspaces

Your CEO Wants You to Do This

If you are assisting your CEO with forecasting tasks, how would you react to a personalized letter, such as this from a software maker:

Mr. P [actual name of your CEO] is interested in this product. What would it be worth to Mr. P if you increase the accuracy of your decisions by 10%? How about 25%? What if you double your effectiveness? Although the future is difficult to foresee, I predict your next “What-If” analysis will result in better insights. Your superiors such as Mr. P will place more confidence in your analysis. You’ll make better decisions. (Brent Green & Associates, 2003)

Such a letter was sent by database marketing firm Brent Green & Associates (BGA) on behalf of the software firm making the forecasting product. The letter was “signed” by the president of this software firm, and BGA selected a list of subscribers to a well-known newsletter targeting financial planners, security investment advisors, security analysts, and brokers. Then BGA matched each prospect on the first list with another list containing names of company presidents obtained from American Business Information, which also included company details. Based on such matched names, each prospect received a letter from BGA bearing the message portrayed above. Some of the letter recipients were enraged at this invasion of privacy, and the president of BGA received a few flaming telephone calls. One caller was livid and threatened to notify the Better Business Bureau and promised that his bank would never buy software from this company. Thus, BGA learned that marrying two or more databases is risky, even if creative overlays offer tempting ways to grab attention (Brent Green & Associates, 2003).

Have You Refilled Your Prescription?

Rite Aid Pharmacy wrote a letter, sponsored by a pharmaceutical manufacturer, to Lisa in California to “remind” her to refill a particular prescription medication that she was taking. Shortly thereafter, Rite Aid called her husband to “remind” him to refill a particular prescription medication that he was taking. The woman felt that Rite Aid was using her prescription records and her family’s private medical information to aggressively market prescription medications. She considered the conduct of Rite Aid unprofessional and complained to the California Department of Consumer Affairs (Privacy Rights Clearinghouse, 1999).

PRIVACY-PERMISSION DYNAMICS

D

With increased dataveillance comes increasingly dataintrusive marketing techniques, such as spam e-mail. Publicly accessible databases and data streams might be easily harvested by aggressive spammers, often for unscrupulous use. In the deluge of unwanted spam and popup windows, legitimate and desired marketing interactions could get sidelined. Permission marketing (Godin, 1999), with its very tightly specified “opt-in” e-communi- cation parameters, and e-commerce Web sites, with strong privacy policies, are belated responses to such developments.

In principle, permissions-based e-commerce seems to take care of privacy invasion problems. In practice, things are more complex. In the evolving privacy-permission dynamics in electronic marketspaces, several contentious issues remain. For example,

•Opt-out permission methods, capable of confusing all but the most vigilant consumers, are used far more frequently than consumer-friendly optin methods.

•The scope of permission—its breadth and depth— is often unclear, and companies can take advantage of this by interpreting permissions in broader and deeper ways that favor them rather than the consumers.

•It is unclear whether permission granted by the customer to a marketer is transferable to a third party that acquires the customer’s information from the marketer.

•Customer permission is very business-specific. What happens if the business is acquired and the customer does not want to grant permission to the acquiring company?

•Conversely, opt-out decisions may not be honored by the new owner of the original customer database, thus invalidating the customer’s initial decision.

CRITICAL SOCIAL THEORY

PERSPECTIVES

Inspired by philosopher of technology Mark Poster (1990b; 1995b), we develop a poststructuralist critique of customer database technology. This approach construes information technology such as databases as configurations of language that produce new and signifi-

cant social, cultural, and political representations. The mode of representation of databases is unique because it transforms the items that databases absorb into specific codes. Increasingly, databases absorb consumers, affecting how they are configured by organizations as “customers” and how organizations relate to these digital customer representations.

We analyze the ideological and cultural work that information technology and databases do in organizations and markets. We do not, however, concern ourselves with the technical and computer-scientific aspects of databases and other information systems.

At the center of our discussion are computerized customer databases. Approaching information technology from a poststructuralist perspective, we call attention to the discursive effects of databases, in Foucault’s sense. Foucault’s understanding of discourse and language is of special relevance for our discussion because of the relation he draws between language and the constitution of the subject. On this view, a new language introduces a new way of constituting cultural objects (e.g., human beings, the organization, customers, suppliers) and social reality (e.g., markets, relationships). Computerized databases constitute such a new language, and their introduction has changed the organizational language for understanding the market in general and customers in particular.

Along with the rise of database technologies and electronically mediated languages, new forms of institutional and organizational power have emerged. Two camps, one with a libertarian and the other with a Marxist outlook, have been commenting forcefully on the implications of this new “mode of information” (Poster, 1990b). The libertarians are mostly concerned about the potential centralization of surveillance power in the hands of state and commercial institutions. On this view, database technology and networked communications are regarded as additional components of the always-growing threat of overbearing government and evermore intrusive corporations. For the libertarians, the biggest fear is that databases brimming with detailed personal information will hasten the end of privacy for citizens and of free choice for consumers (Dholakia & Zwick, 2001).

Marxists are more concerned about the contribution of computer networks and database technology to the dominance of corporations over the working class. Because information is not equally available to all, this camp maintains that capitalists can seize information technologies to further monopolize control over the means of production. In the information economy, such control includes knowledge workers as well as manual workers. Such a position suggests that management mobilizes the rhetoric of efficiency to obtain workers’ acceptance for erecting a fully transparent organization, where

Dataveillance and Panoptic Marketspaces

all production processes are always under surveillance, every member is monitored, and his or her contribution is assessed constantly.

Both viewpoints have much to offer for our understanding of the impact of information technology on personal freedom, consumer sovereignty, and the power of the worker. Yet, they fail to grasp the cultural innovations brought about by the integration of information technologies, such as databases, into existing political, economic, and social institutions (Poster, 1990b, 1995b; Sotto, 1997). The problem of information technology’s cultural effects escapes Marxists and liberal thinkers because they theorize the social field primarily as one of action, minimizing the importance of language. Yet, databases are made up of symbols in data fields. They embody a specific mode of representing the world—what Bolter (2001) calls “numeric inscription.” As Poster (1995) puts it, “One does not eat them, handle them, or kick them, at least one hopes not. Databases are configurations of language; the theoretical stance that engages them must take at least this ontological fact into account.”

The use of information technology to transform citizens, workers, and consumers into digital representations is relevant because it enables new forms of surveillance, control, and management of these populations. Before the arrival of the electronic database and the widespread use of electronic and digital transaction formats in the marketplace, the consumer was not subjected to a permanent and ubiquitous regime of surveillance and observation. The “massified” consumer was able to slip in and out of markets, easily and anonymously. The database, however, is like an air traffic control tower that presents a screen of almost total visibility. On this visual screen, the individual customer is distributed relative to other customers, according to a system of similarities and differences operating in two steps: first, the encoding, and second, the interpretation of details of personal attributes and conduct. The customer is no longer “lost in the fleeting passage of time, space, movement, and voice but identifiable and notable” (Rose, 1988). Within the regime of norms and requirements that make up this perceptual surface, each customer—encoded according to a clearly defined language of data fields and algorithms—can now be placed in relation to other customers. The electronic and digital transaction space between organizations and their customers becomes a map, which enables the emergence of what was previously imperceptible and below the threshold of description: the individual customer.

Thus, with the increases in customer database profile sizes, the threshold for detecting ordinary individuality has been continuously lowered. Therefore, it might be useful to think about the mode of functioning of the

Dataveillance and Panoptic Marketspaces

customer database as a representational or discursive machine that identifies and harnesses human difference. In this respect, customer databases are conceptually closely related to Foucault’s notion of the Panopticon. Foucault developed this idea from Jeremy Bentham’s late 18th-century description of an architectural structure that depicted a circular-type prison of individual cells, arranged around the perimeter. A tower rises in the center of the prison, allowing a single warden to see any cell at any time. Foucault regarded the Panopticon as a model for the transition from a traditional to a modern form of society. For him, “this enclosed, segmented space, observed at every point, in which individuals are inserted in a fixed place, in which the slightest movements are supervised, in which all events are recorded, in which an uninterrupted work of writing links the center and periphery…in which each individual is constantly located, examined, and distributed among the living beings” becomes the place where the new technology of observation and recording creates “new truths” about the individual. The Panopticon is therefore less a force of repression and oppression than an inscription and discursive fabrication of individual.

The Panopticon, as a metaphor for our surveillance society, denotes an accumulation and centralized knowledge in the hands of the technology holder. In these databases, consumers take discernible shape for data mining purposes, where they are configured as datalinguistic constructs that can be identified and acted upon strategically. Today, the recording of all events and the databases they generate constitute a Superpanopticon, a system of surveillance without walls, windows, towers, or guards (Poster, 1990). Technological advances, however, are only part of the story. The consumer has become accustomed to surveillance and even to participating in the process of data accumulation. Social Security numbers, credit cards, library cards, drivers’ licenses, and loyalty cards—the consumer must apply for them, use them continuously, have them ready at all times, and verify their validity by cross-referenc- ing them with other documents. Each transaction is recorded, encoded, and added to the database. The circle of seamless and continuous dataveillance has been closed.

Thus, though consumers participate in the formation and population of their own data records by committing simple consumption acts, they do not completely control this (in)formation. In fact, with online tracking, powerful recoding and filtering technologies, and more frequent selling and exchanging of customer records, consumers are no longer able to always decide what kind of information about them is stored, categorized, manipulated, exchanged, and acted upon by whom, when, and where. What is more important, matters of

control over one’s personal information are complicated

by the emergence of new information technologies. Radio D Frequency Identification (RFID) technologies, for example, can be used for smart tags. RFID tags are used to

track assets, manage inventory, and authorize payments, and they increasingly serve as electronic keys for everything from automobiles to secure facilities. The European clothier Benetton was among the first consumer product manufacturers to equip merchandise such as pants and sweaters with RFID tags by weaving them into the traditional garment tag. Though the tag does not directly store information about the person wearing the Benetton product, it nevertheless has the potential to communicate much about him or her. Consider a person wearing Benetton pants equipped with an RFID tag during a shopping trip in a downtown mall. Inconspicuous receivers placed in stores and public spaces could continuously communicate with the pant’s smart tag, eventually painting a detailed picture about the person’s shopping trajectory.

In sum, databases form the basis of the panoptic marketspace. The act of consumer individualization—based on the workings of the database—transforms the ways in which inscriptions of human individuality can be produced, ordered, accumulated, and circulated. In its essence, database-driven information technologies “provide a technique of visualization and inscription of individuality which objectify their subjects by inscribing their differences from one another” (Rose, 1988, p. 195). The ways in which consumers are dataveilled, fabricated as data objects, and acted upon to gain control over their behaviors need to be understood in order to develop strategies to counteract the imbalance of panoptic power. The accumulation and centralization of knowledge about the consumer in customer databases make the database the site where customers are constituted as culturally relevant objects of analysis, knowledge, and managerial action. Any form of political action that aims to curtail the power of the panoptic marketspace must take place at the level of the database.

POLICY IMPLICATIONS

With customer databases growing and becoming ubiquitous, consumers are becoming increasingly known and acted upon as digital representations. Hence, the constitution of the customer as epistemological entity takes place within the code of the database. In addition, it is increasingly impossible for consumers to prevent dataveillance by marketers or to participate in the formation of their own data profiles. Therefore, actions that go beyond the generation and application of individual information externalization strategies are needed. Indeed, we argue that consumers must be given direct access to customer databases to ensure that they regain a viable

voice in the process of their constitution as digitized customers.

Always at the leading edge of database marketing, Amazon.com has come to understand that constituting valuable and strategically superior digital consumer identities cannot be accomplished without customers’ direct access to its databases. After recognizing return customers via sign-ins and cookies, Amazon.com offers them dozens of product recommendations. After each recommendation, the hyperlink “Why was I recommended this?” is offered. Clicking on this hyperlink offers a glimpse into the section of the Amazon database that stores previous, correlated purchases of the customer. The customer then has options of clicking on “Not Interested” in that specific recommendation or to intensify or dilute the recommendation algorithm by tweaking the intensity of preference for the previously purchased, correlated items on a 5-point scale. In effect, Amazon.com lets consumers take partial control of the constitution of their customer identity by allowing them to access and alter the code and content of the database.

Similarly, in Canada the privacy commissioner of the Province of Ontario has announced an initiative that will open up all government databases to its citizens, enabling them to make changes to the data profile currently stored there. Lyotard (1984) proclaimed some time ago that we must “give the public free access to the memory and data banks” (p. 67). Twenty years later, his call may finally be heeded.

CONCLUSION

Dataveilled consumers constitute digitized customer representations in vast databases—this is the reality of contemporary business in advanced economies. Incentives and sanctions are in place to ensure that consumers participate—willingly or reluctantly—in this ongoing accumulation of data.

For business, consumers, and public policymakers, there are serious challenges ahead. In panoptic, digitized customer databases, there are many possibilities of privacy invasion, misrepresentation, identity theft, and other social ills. Therefore, from private business-strategic perspectives as well as enlightened public policy perspectives, it is necessary take actions that go beyond the generation and application of individual information externalization strategies. Customer databases will not go away—they will only increase in size and sophistication with the advent of new technologies such as RFID tags and biometrics. Therefore, enlightened policies should ensure that consumers have direct access to customer databases and that they have some say in the processes of their constitution as digitized customers.

Dataveillance and Panoptic Marketspaces

REFERENCES

Bolter, J. D. (2001). Writing space: Computers, hypertext, and the remediation of print. Mahwah, NJ: Erlbaum.

Brent Green Associates. (2003). http://bgassocia tes.com/ essay2.htm

Clarke, R. (1988). Information technology and dataveillance.

Communication of ACM, 31(5), 498-512.

Dholakia, N., & Zwick, D. (2001). Privacy and consumer agency in the information age: Between prying profilers and preening Webcams. Journal of Research for Consumers, 1(1), Article 3. Retrieved from http:// www.jrconsumers.com

EPIC. (2003). http://www.epic.org/privacy/airtravel/ nwa_comp.pdf

Godin, S. (1999). Permission marketing: Turning strangers into friends, and friends into customers. New York: Simon & Schuster.

Ha, M. (2003, April 21). Consumer groups try to push big brother insurer out of passenger seat. National Underwriter (Property & Risk Management Edition), 107(16).

Kroker, A., & Weinstein, M. A. (1994). Data trash: The theory of the virtual class. Montreal, Canada: New World Perspectives.

Lyotard, J.-F. (1984). The postmodern condition: A report on knowledge: Vol. 10 (G. Bennington & B. Massumi, Trans.). Minneapolis: University of Minnesota Press.

Poster, M. (1990a). Foucault and databases. Discourse, 12(2), 110-127.

Poster, M. (1990b). The mode of information. Chicago: The University of Chicago Press.

Poster, M. (1995a). Databases as discourse, or electronic interpellations. In P. Heelas, S. Lash, & P. Morris (Eds.), Detraditionalization (pp. 277-293). Oxford, UK: Blackwell.

Poster, M. (1995b). The second media age. Cambridge, MA: Polity Press.

Privacy Rights Clearinghouse. (1999). http://www.pri vacyrights.org/

Rayport, J. F., & Sviokla, J. J. (1994). Managing in the marketspace. Harvard Business Review, 72(6), 141-150.

Dataveillance and Panoptic Marketspaces

Rose, N. (1988). Calculable minds and manageable individuals. History of the Human Sciences, 1(2), 179-200.

Sotto, R. (1997). The virtual organization. Accounting, Management and Information Technology, 7(1), 37-51.

Zwick, D., & Dholakia, N. (2004). Whose identity is it anyway? Consumer representation in the age of database marketing. Journal of Macromarketing.

KEY TERMS

Dataveillance: The monitoring of people by digital representations in electronic databases created and managed by information technologies.

Marketspace: Electronic transaction methods, or electronic markets, in which businesses and consumers interact.

M-Commerce: Mobile commerce, transactions on mobile phones. Examples include, in addition to voice communications, SMS messages, game downloads, stock market quotes, and the like.

INTRODUCTION

Integrity constraints of various kinds must be observed when creating or updating a database in order to preserve the semantics and the quality of stored data (Elmasri & Navathe, 2000). Within the scope of geographic applications, integrity assurance requires special attention from the designer, since most geographic applications use data that depend on spatial relationships (Egenhofer & Franzosa, 1991), thereby requiring the specification of spatial integrity constraints.

In the traditional database approach, there is a relationship between conceptual, logical, and physical design, in which, through mapping operations, constraints that are identified in the conceptual schema are inherited and transformed into implicit constraints expressed by the data definition language (DDL) or into explicit constraints coded in the application programs (Elmasri & Navathe, 2000). This relationship also exists in spatial information systems; therefore, spatial constraints can be likewise identified and implemented. However, even though there is a very active research area interested in the design of robust and efficient spatial databases, there are still shortcomings with respect to spatial integrity constraints (Borges, Davis & Laender, 2002; Plumber & Groger, 1997). This happens mostly because a simple geometric modification in a single object in a spatial database may generate the need to check for possible integrity violations throughout many object classes, using computationally intensive geometric and topologic algorithms.

Most spatial integrity constraints are, in fact, semantic integrity constraints applied to the spatial representation of objects and to the relationships among object instances that are based on spatial representations. In order to be able to adequately represent such representa-

tions and relationships in geographic applications design, tools that are more specific and capable of capturing the semantics of geographic data, offering higher abstraction mechanisms and implementation independence (Borges, Davis & Laender, 2001; Câmara, 1995) are required. From geographic application database schemas, developed using an adequate data model, representations and spatial relationships can be extracted; thus, spatial integrity constraints can be specified.

This paper focuses on the types of spatial integrity constraints that derive from spatial data modeling constructs, as a part of spatial databases design. OMT-G (Borges et al., 2001), an object-oriented data model for geographic applications, is used to illustrate the concepts involved in the definition of such constraints.

BACKGROUND

Every data model has a set of built-in constraints associated with its constructs (Elmasri & Navathe, 2000). Accordingly, the OMT-G model allows several spatial integrity rules to be derived from its primitives. These rules constitute a set of constraints that must be observed in the operations that update a geographic database. Ideally, these rules should be implemented by the spatiallyenabled database management system (DBMS) in its data definition language (DDL), along with the necessary expansions to the data manipulation language (DML) and the implementation of geographic data types and access methods. However, the approach employed by current commercial products is rather different. Since current spatial DBMSs do not implement spatial integrity constraints, the task of ensuring consistency ends up in the application, usually developed as a geographic information system (GIS). In general, GIS tools allow inconsistent

information to enter the database; later on, a range of correction functions is used to “clean up” the data, verifying its consistency.

Regardless of whether the implementation of integrity constraints is to be performed by the DBMS or by the GIS, the required constraints can be determined at the conceptual level of spatial databases design. In order to show how this can be done, we will use conceptual schema primitives from the OMT-G model (Borges et al., 2001), showing how the semantics embedded in the primitives leads to the spatial integrity constraints. Such an exercise can be performed on any other spatial data model.

SPATIAL INTEGRITY CONSTRAINTS AND THE OMT-G MODEL

OMT-G is a conceptual data model based on the primitives of the UML class diagram (Rational, 1997), enhancing it with geographic primitives. It is based on three main concepts: classes, relationships, and spatial integrity constraints. Classes and relationships are the basic primitives that are used to create application schemas with OMT-G. From these primitives, spatial integrity constraints can be obtained, as presented next.

Classes, in OMT-G, represent the three main groups of data (continuous, discrete, and non-spatial) that can be found in geographic applications. Non-spatial data are represented using conventional classes, which can relate to spatial objects but have no geometric or geographic properties. Georeferenced classes describe a set of objects that have spatial representation and are associated to features on Earth (Câmara, 1995). Assuming the fields and objects view (Frank & Goodchild, 1990), georeferenced classes are specialized into geo-field and geo-object classes. Geo-field classes represent objects and phenomena that are continuously distributed over the space, corresponding to variables such as soil type,

relief, and mineral contents. Geo-object classes represent individual, particular geographic objects, which can usu- D ally be traced back to real-world elements, such as build-

ings, rivers, and trees. The notation employed by OMT- G is shown in Figure 1.

There are five geo-field descendant classes in OMT- G: isoline, planar subdivision, tesselation, sampling, and triangular irregular network (Figure 2). From the semantics involved in the concept of geo-fields and from the specific definition of these classes, some spatial integrity rules can be deduced (Table 1).

Geo-object classes are classified into geo-object with geometry classes, representing objects which have only geometric properties (points, lines, and polygons), and geo-object with geometry and topology, which represents objects which also have topological connectivity properties, and are thus specifically suited to the representation of spatial network structures. Topological properties are present in objects that are either nodes or arcs in a graph-theoretic approach, thereby forming Node, Unidirectional Line, and Bidirectional Line classes (Figure 3). Geo-objects with geometry and topology are not subject to a set of integrity constraints by themselves, but their use is conditioned to the existence of network relationships (Table 4).

The geometric concepts used in the definition of points, lines (including lines with a topological role), and polygons lead to some integrity constraints. The geometric definitions adopted in OMT-G admit the existence of geo-objects that are formed by several polygons, establishing one of them as the “basic” polygon and considering the others as islands or holes. Polygons that are composed of multiple parts (or polygonal regions) are important, since there is no guarantee that the results of traditional operations, such as buffer creation, union, intersection, and difference between simple polygons, are always formed with simple polygons. Constraints regarding lines and polygons are presented in Table 2.

Figure 1. Graphic notation for the basic classes

Figure 2. Geo-field classes

Planar subdivision

Figure 4. Relationships

(a) Sim ple association

(b) Spatial relationship

(c) Arc-node network relationship

Highway

Highway system

(d) Arc-arc network relationship

OMT-G includes three basic relationship types (simple associations, spatial relations, and network relations) (Figure 4), along with object-oriented relationships (generalization/specialization, aggregation and conceptual generalization).

Table 3. Spatial relationship integrity rules

Simple associations represent structural relationships between objects of different classes, conventional as well D as georeferenced. Integrity constraints corresponding to simple associations are, thus, essentially the same constraints usually implemented in conventional databases.

Spatial relations represent topologic, metric, ordinal, and fuzzy relationships. Some relations can be derived automatically, from the geometry of each object, during the execution of data entry or spatial analysis operations. Others need to be specified by the user, in order to allow the system to store and maintain that information. OMT- G considers a set of five basic spatial relations between georeferenced classes, from which all others can be derived (Clementini, DiFelice & Oosterom, 1993): touch, in, cross, overlap, and disjoint. Since sometimes a larger set is required, due to cultural or semantic concepts that are familiar to the users, OMT-G also includes relations such as adjacent to, coincide, contain, and near, which are special cases of one of the five basic relations but deserve special treatment because of their common use in practice. Alternatively, the set of basic spatial relations derived from the 4-intersection matrix (disjoint, contains/inside, covers/covered by, meet, overlap, equal), or the spatial relations derived from the 9-intersection matrix, can be employed (Egenhofer & Franzosa, 1991; Egenhofer & Mark, 1995).

Considering these spatial relationship types, some spatial integrity rules can be established (Table 3). These rules are formulated using a notation commonly found in computational geometry, in which objects are indicated