Interoperating Geographic Information Systems M F Goodchild, M J Egenhofer, R Fegeas and C A

Kottman (eds). Kluwer.

0 CHANGING FOCUS ON
INTEROPERABILITY
IN INFORMATION SYSTEMS:
FROM SYSTEM, SYNTAX, STRUCTURE
TO SEMANTICS
Amit P. Sheth

1. Introduction

Interoperability has been a basic requirement for the modern information systems
environment for over two decades. How have key requirements for interoperability
changed over that time? How can we understand the full scope of interoperability
issues? What has shaped research on information system interoperability? What key
progress has been made? This chapter provides some of the answers to these
questions. In particular, it looks at different levels of information system
interoperability, while reviewing the changing focus of interoperability research
themes, past achievements and new challenges in the emerging global information
infrastructure (GII). It divides the research into three generations, and discusses some
of achievements of the past. Finally, as we move from managing data to information,
and in future knowledge, the need for achieving semantic interoperability is
discussed and key components of solutions are introduced.
Data and information interoperability has gained increasing attention for several
reasons, including:
• excellent progress in interconnection afforded by the Internet, Web and
distributed computing infrastructures, leading to easy access to a large number
of independently created and managed information sources of broad variety;
• increasing specialization of work, but increasing need to reuse and analyze data,
leading to creation of information and knowledge, and their subsequent reuse
and sharing.
1
2 Amit P. Sheth

Any attempt to give a broad survey of information systems interoperability is

difficult and complex for several reasons, including different levels of requirements,
the variety of approaches, the number of technical areas involved, the large literature,
and the large number of example systems. For summative and pedagogical reasons,
we make some broad observations and avoid paying attention to many exceptions.
We propose to view information system evolution in the context of
interoperability as occurring in three generations: Generation I which covers the
period to roughly 1985, Generation II which covers the period of a decade through
1995, and Generation III which covers a period yet to be bounded since 1996. Our
discussion primarily draws from database and information system research, while
noting that relevant work can also be found in information retrieval, knowledge-
based systems and AI, multimedia, and other disciplines. Also very little attention is
given in this chapter to several application domains, for example, geographic
information systems (Goodchild et al. 1997), in which interoperability has received
significant attention.
One of the enduring approaches to studying the key interoperability issues in
multidatabases and federated databases of the first generation has been to use the
fundamental dimensions of distribution, heterogeneity, and autonomy (Sheth and
Larson 1990). The same dimensions also provide a good starting point for studying
interoperability issues in the subsequent generations. After a brief discussion of the
distribution and autonomy dimensions, we will focus on the heterogeneity dimension
as we study interoperability issues in the three generations.

1.1. Distribution

The scope of interoperability during the first generation was primarily departmental
and almost always within a company. Usually, the multidatabase systems involved
just a few databases and computer nodes, either connected point-to-point or in a local
area network. With the significant impact of the Internet and advent of the Web, the
scope of interoperability during the second generation has been enterprise-wide as
well as inter-enterprise. It was not unusual to find tens of computers and data
repositories involved in a second-generation system. In the third generation, with
significant improvements in communication technology, global information
infrastructure, and distributed computing infrastructure, the dimension of distribution
of data has achieved a very broad scope—from a single system to global. As the
distributed nature of data and information is often hidden from the end users, the
system developers face several new challenges. A few of the noteworthy challenges
involve increasing use of large amounts of data and information sources—
particularly involving visual data, use of a wide variety of communication modes
with a variety of bandwidths, and a larger optimization space involving varying
capabilities of the component systems. Compared to the first generation systems, the
issue of optimization has received less attention in the second generation.

1.2. Autonomy

The organizational entities that manage different information sources (including

database systems, DBSs) are often autonomous. Those who control an information
Changing Focus on Interoperability in Information Systems 3

source are often willing to let others share the data only if they retain control. Thus it
is important to understand the aspects of autonomy and how they can be addressed
when a database system participates in a federation or shares its data with new users
or applications.
Let us look at a classification of autonomy issues in the context of federated
database systems (adapted from Sheth and Larson 1990), noting that these can be
adapted to other architectures by considering the various types of information
sources and information system components involved. A component participating in
a federation may exhibit several types of autonomy, including design,
communication, association and execution.
Design autonomy refers to the ability of a component to choose its own design
with respect to any matter, including
• the data or information being managed (i.e., the Universe of Discourse or
domain),
• the representation (data model, query language) and the naming of the data
elements (or the ontology used),
• the conceptualization or semantic interpretation of the data (or the context),
• constraints used to manage the data,
• the functionality of the system,
• association and sharing with other systems (see association autonomy below),
and
• the implementation (e.g., record and file structures, concurrency control
algorithms).
Communication autonomy refers to the ability of a component to decide whether
to communicate with other components. A component with communication
autonomy is able to decide when and how it responds to a request from another
component. Execution autonomy refers to the ability of a component to execute local
operations without interference from an external entity and to decide the order in
which to execute external operations. Thus, an external system cannot enforce an
order of execution of the commands on a component with execution autonomy.
Execution autonomy implies that a component can abort any operation that does not
meet its local constraints and that its local operations are logically unaffected by its
participation in a federation. Furthermore, the component does not need to inform an
external or federated system of the order in which external operations are executed
and the order of an external operation with respect to local operations. Operationally,
a component exercises its execution autonomy by treating external operations in the
same way as local operations.
Association autonomy implies that a component has the ability to decide whether
and how much to share its functionality (i.e., the operations it supports) and
resources (i.e., the data it manages) with others. This includes the ability to associate
or disassociate itself from the federation and the ability of a component to participate
in one or more federations. Several first-generation systems in the database area paid
significant attention to the autonomy issue because they also attempted to support
4 Amit P. Sheth

updates. In comparison, few second-generation systems have considered update

issues. Although autonomy of components has been assumed in almost all systems,
there has been virtually no attention paid to the challenges posed by the various
aspects of the autonomy dimension.

1.3. Heterogeneity

Many types of heterogeneity are due to technological differences; for example,

differences in hardware, system software (e.g., operating system), and
communication systems. Researchers and developers have been working on
resolving such heterogeneity for many years. Figure 1 shows one perspective on
heterogeneity. Focusing on the crucial dimension of heterogeneity and
corresponding solutions leads us to discuss different levels of interoperability—
system, syntax, structure, and semantic. In this classification, we consider differences
in machine-readable aspects of data representation, also referred to as formatting, to
be relevant to syntactic heterogeneity. We consider representational heterogeneity
that involves data modeling constructs to be relevant to structural interoperability.
Schematic heterogeneity that particularly appears in structured databases is also an
aspect of structural heterogeneity.
Changing Focus on Interoperability in Information Systems 5

Figure 1. Heterogeneity in information systems

During the second half of the 1970s, we saw the ability to deal with hardware,
operating systems, and communications heterogeneity; although with evolution in
each of these, new issues have to be continuously addressed. During the 1980s, we
saw significant progress in managing heterogeneity and support interoperability or
integration in environments with structured databases and traditional database
management systems (DBMSs). There is a large body of work during the first
generation in dealing with heterogeneity associated with data models or schematic
issues, DBMSs including query languages, concurrency control, commit and
recovery, etc.
During the 1990s, the emergence of distributed computing, middleware
technology, and standards has allowed us to increase focus on the heterogeneity that
is intrinsic to data (or media). This has particularly supported syntactic and structural
interoperability, and allowed us to address issues at the information level. As the
future information system increasingly addresses the information and knowledge
level issues, it will increasingly require semantic interoperability. Semantic
interoperability requires that the information system understands the semantics of the
6 Amit P. Sheth

user’s information request and those of information sources, and uses mediation or
information brokering to satisfy the information request as well as it can.
The remainder of this chapter provides an overview of the three generations of
systems, with emphasis on the heterogeneity dimension of support for different
levels of interoperability. Table 1 provides an overview of the three generations in
terms of a variety of criteria.

Generation I Generation II Generation III

Level of system, data system, data, System, data,
interoperation information information, knowledge
concern (new (incl. social), process
emphasis
underlined)
Types of system (computer syntax (data types and semantic (increasingly
interoperability system and formats), structure domain-specific)
emphasized communication); (schematic, query
limited aspects of languages and
syntax and structure interfaces)
(data model);
transparency of
location, distribution,
replication, data
models
Dominant multidatabases or federated information mediator, information
interoperability federated databases systems, mediator brokering
architecture
Scope of system handful of tens of systems on a enterprise-wide and
interoperability interconnected LAN, databases and text global scope
computers and repositories
databases
Software and terminal access, client-server (three tier); network, distributed, and
information point-to-point; also mobile
system mainframes and
minicomputers with
architecture
remote access, client-
server (two-tier);
Communication proprietary (IBM TCP/IP, http, CORBA Internet/Web/Java,
infrastructure domination), TCP/IP distributed object
on which system management, component,
but increasingly higher-
interoperability
level such as multi-agent,
solutions are mobile
built
Types of data structured databases structured databases, all forms of digital media
and files text repositories, semi- with increasing support
structured and for visual/spatio-
structured and data in temporal/
generic (e.g., SGML, scientific/engineering
HTML) and domain data;
specific formats
Dominant Relational and E-R object-oriented component-based; multi-
information modal
source/system
model
Changing Focus on Interoperability in Information Systems 7

Data/ structural and data understanding of a comprehensive use of

information model, data variety of metadata, metadata, increasing
interoperability representation comprehensive emphasis on semantics
understanding of and ontology supported
approaches
schematic heterogeneity approaches
Interoperability data-level schematic and metadata- multiple ontologies,
techniques relationships, level relationships, information or semantic
(representative common/canonical wrappers, extractors, level relationships,
data models, single ontology, context, media-
samples)
mappings, database metadatabase, independent information
exchanges, remote schematic correlation, inter-
database interfaces, heterogeneity, ontological relationships,
query multidatabase metadata consistency
transformations, consistency, mediators
schema translation,
schema integration
Key human data(base) software developers to domain experts for
roles in administrators or generate wrappers and ontologies and for
supporting experienced users, mediators (with some generating information
knowledgeable data toolkits) involving data correlations
interoperability
structures and level issues
models, software
developer written
access programs
Access options database query keyword-based attribute multimedia views; visual
language (SQL) for and (limited) content- interfaces; information
structured databases, based access, (limited) requests that are media-
keyword accesses for ontology-based access, independent, multi-
textual data/files ontology based, context-
sensitive and domain-
specific
A few integration of digital library, digital earth,
representative business databases or integrated access to environmental
applications public databases heterogeneous data for a phenomena, multi-step
software team and multi-modal
intelligence analysis
One Find a four star Find flowers suitable for Find a block of land with
representative restaurant with less winter gardens that look urban land cover and
complex query than $25 average cost like this <image> with a moderate relief and
that serves soft smell (a keyword-, population greater than
Mediterranean food attribute-, and content- 5000 and area greater
in Richmond (a based query on text and than 1000 sq ft suitable
multidatabase query image data repositories) for a strip mall (a query
on distributed with terms whose
structured databases) meanings are understood
by the system, and may
involve multi-step
processing against multi-
modal data)
Research ADDS, DDTS, GARLIC, Harvest,
prototypes Interbase, Mermaid, HERMES,
MRDSM, Multibase, InfoHarness/Visual-
Omnibase, see Sheth Harness, Information
and Larson (1990) Manifold, InfoSleuth,
for more examples RUFUS, SIMS,
TSIMMIS, …
Products UniSQL/M AdaptX/Harness
(Companies) (UniSQL), Mermaid (Bellcore),
8 Amit P. Sheth

(Data Integration), (Junglee), TIE

DataJoiner (IBM), (Tesserae),
OmniConnect (Excalibur)
(Sybase)

Table 1: An overview of three generations of introperability R&D

2. First generation

By the 1980s, corporations had amassed large amounts of data in different

departments to serve different applications, and on computers with different
hardware and software (including DBMSs). The mantra of “data is a corporate
resource” drove needs for exchanging and sharing the data between departments and
within enterprises. Perhaps the most representative work in this generation occurred
in the context of multidatabases (Litwin et al. 1982), or federated database systems
(Heimbigner and McLeod 1985; Sheth and Larson 1990). We will focus on some of
the key understandings gained during this period (Drew et al. 1993; Elmagarmid
1992; Elmagarmid and Pu 1990; Elmagarmid et al. 1998; Hsiao et al. 1993;
Kambayashi et al. 1991; Kim 1995; Ram 1991; Schek et al. 1993; Sheth 1987).
Much of the emphasis during this generation was on achieving system
interoperability, in particular by addressing the heterogeneity due to differences in
DBMSs, with some work on syntactic heterogeneity and schematic heterogeneity as
appropriate to the structure of databases. Correspondingly, the emphasis was on data
management and data (as opposed to information or knowledge). Let us discuss these
briefly, as good understanding on these issues has been achieved.
Each DBMS has an underlying data model used to define data structures and
constraints. Both representation (syntactic as well as structural issues including
constraints) and language aspects can lead to heterogeneity. Following are some of
the issues that received significant attention from the database and information
systems research community.
Differences in structure: Different data models provide different structural
primitives (e.g., the information modeled using a relation or table in the relational
model may be modeled as a record type in the network database model.). If the two
representations have the same information content, it is easier to deal with the
differences in structures. For example, address can be represented as an entity in one
schema and as a composite attribute in other schema. If the information content is
not the same, it may be very difficult to deal with the difference. As another
example, some data models (notably semantic and object-oriented models) support
generalization (and property inheritance) while others do not.
Differences in constraints: Two data models may support different constraints. For
example, the set type in a schema based on a network database model may be
partially modeled as a referential integrity constraint in a relational schema.
However, a network database model supports insertion and retention constraints that
are not captured by the referential integrity constraint alone. Triggers (or some other
mechanism) must be used in relational systems to capture such semantics.
Changing Focus on Interoperability in Information Systems 9

Differences in query languages: Different languages are used to manipulate data

represented in different data models. Even when two DBMSs support the same data
model, differences in their query languages (e.g., QUEL and SQL), or different
versions of SQL supported by two relational DBMSs) could contribute to
heterogeneity.
Differences in the system aspects of the DBMSs: Examples of system-level
heterogeneity include differences in transaction management primitives and
techniques (including concurrency control, commit protocols, and recovery),
hardware and system software requirements, and communication capabilities.
A number of prototypes, including some of the well-known systems such as
Multibase, Mermaid, DDTS, ADDS, or MRDSM addressed a variety of technical
issues such those listed above. However there has been very little commercialization
resulting from these efforts, and the success of commercial systems has been limited.
Reasons are both technical and business (see Sheth 1995 for a discussion).
Elmagarmid et al. (1998) provide a more recent review of technical work during this
generation.
10 Amit P. Sheth

Figure 2. Schema architecture for a federated database system

Figure 2 shows how some of the distribution, autonomy, and heterogeneity issues
for integrating component databases can be handled (see Sheth and Larson 1990 for
more details). Briefly, the local schemas can represent data in the data model of
respective DBMSs. To be able to compare the data objects modeled in different data
models, one has either to perform direct and pairwise comparison—something like
comparing apples and oranges—or to convert the schemas to a common or canonical
Changing Focus on Interoperability in Information Systems 11

model, preferably with an expressive power exceeding that of models for component
databases, and then compare objects. Defining export schemas allows handling of
one aspect of autonomy. Integration of export schemas into federated schemas
allows for integrated or uniform access to objects managed by multiple component
databases. Defining external schemas allows for handling additional types of
heterogeneity.
DBMS-level heterogeneity covers only a small set of heterogeneity related to
structure databases. Figure 3 shows one classification of a variety of conflicts related
to achieving interoperability among or integration of multiple databases managed by
traditional DBMSs (Kim et al. 1993; Sheth and Kashyap 1993).

Figure 3. A classification of conflicts in structured databases

We believe that some of the lasting contributions of research in this generation

are:
• understanding of the data models for structured databases, and the techniques to
deal with data modeled with different models (so-called schema translation
issue);
• techniques for schema integration (Batini et al. 1986) and understanding the
complexity (and at times the futility) of performing integration involving real-
world schemas (Sheth 1995);
• understanding of and finding ways to deal with schematic or representational
heterogeneity, and the early realization of the distinction between
schematic/representational issues and semantic issues in dealing with
relationships and possible integration of data managed by different sources (or
DBMSs; Drew et al. 1993; Meersman and Mark 1997; Sheth 1991; Sheth and
Kashyap 1993);
12 Amit P. Sheth

• issues of managing consistency among data managed by different databases, and

corresponding research in advanced transaction models (Elmagarmid 1992) and
multidatabase transactions execution (Georgakopolous and Rusinkiewicz 1994)
The predominant architectural framework during the first generation systems was
that of the federated databases (Heimbigner and McLeod 1985; Sheth and Larson
1990). A federated database system (FDBS) architecture consists of a schema
component, such as the one shown in Figure 2, interspersed with processors (such as
transforming processors to help with heterogeneity, filtering processors to help with
autonomy, and constructing processors to help with integration of distributed data).
Two broad subcategories of FDBS architectures emerged:
• a loosely-coupled architecture that provided for a more dynamic or flexible
federation (where it is easier for a component to join or leave), usually with no
support for updates, but with reliance on more end user involvement and less
transparency of resources; and
• a tightly-coupled architecture that provided for more stable federation, with
ability to support updates and distributed transactions as well as better
integration of data managed by components, but requiring more systems
administrator involvement for the tasks such as schema integration
One of the major issues that this generation addressed but that subsequent
generations have yet to consider is that of updates, and correspondingly data
consistency and use of transaction mechanisms.

3. Second generation

During the second generation two very important trends brought extraordinary
opportunities for interoperability and exploitation of data: (a) proliferation of a
variety of data—from structured database, and semi-structured data, to digital media,
including visual media (Gupta and Jain 1997), and (b) spread of the Internet and
emergence of the Web. Applications such as digital libraries (Paepcke et al. 1998)
and electronic commerce provided the context of interoperability.
Some of the key trends and achievements of this generation are (Papazoglou and
Schlageter 1998; Sheth and Klas 1998):
• technology for dealing with heterogeneity of systems, data, and representational
levels;
• support for a broader variety of data—not just structured databases, but also text,
semi-structured, and unstructured (including image and video) data;
• use of a broad variety of metadata to support interoperability and integration;
and
• use of knowledge representation and reasoning, especially for handling
terminological differences.
Changing Focus on Interoperability in Information Systems 13

3.1. Metadata

Metadata are usually defined as data about data. Often they is more than that,
involving information about data as they is stored or managed, and revealing partial
semantics such as intended use (i.e., application) of data. This information can be of
broad variety, meeting if not surpassing the variety in the data themselves. Metadata
can be regarded as an extension (albeit a significant one) of the concept of the
schema in structured databases. They may describe, or be a summary of the
information content of the individual databases in an intentional manner. They
typically represent constraints between the individual media objects that are implicit
and not necessarily represented in the databases themselves. Some metadata may
also capture content-independent information like location and time of creation.
Examples of what we consider media types are structured data (data in relational or
object-oriented databases), textual data (of different formats, such as Word files,
source code, etc.), images (of possibly different modalities such as X-Ray, MRI
scan), audio (of possibly different modalities such as monaural, stereophonic), and
video. Sheth and Klas (1998) give an extensive discussion on types of metadata and
their applications in managing and exploiting various digital media.
The criterion we use to classify metadata (Kashyap et al. 1995) is the extent to
which they are successful in capturing the (data and information) content of the
information asset (also called artifact or document in different contexts) represented
in various media types. The level of abstraction at which the content of the assets is
captured is very important. We believe that to capture the semantic content (i.e., at a
level of abstraction closer to that of humans), it is important for the metadata to
model application domain-specific information. Metadata descriptions present two
advantages:
1. They enable the abstraction of representational details such as the format and
organization of data, and capture the information content of the underlying data
independent of representational details. This represents the first step in reduction
of information overload as intentional metadata descriptions are in general an
order of magnitude smaller than the underlying data.
2. They enable representation of domain knowledge describing the information
domain to which the underlying data belong. This knowledge may then be used
to make inferences about the underlying data. This helps in reducing information
overload as the inferences may be used to determine the relevance of the
underlying data without accessing the data.
One of several classifications of metadata (Kashyap et al. 1995) is as follows
(also see Boll et al. 1998; Lagoze et al. 1996):
Content-independent metadata: This type of metadata captures information that
does not depend on the content of the asset with which it is associated. Examples of
this type of metadata are location, modification date of a document and type of
sensor used to record a photographic image. There is no information content
captured by these metadata but these might still be useful for retrieval of assets from
their actual physical locations and for checking whether the information is current or
not.
Content-dependent metadata: This type of metadata depends on the content of the
asset it is associated with. Examples of content-dependent metadata are size of a
14 Amit P. Sheth

document, maximum number of colors, number of rows, number of columns of an

image. Content-dependent metadata can be further subdivided as follows:
Direct content-based metadata: This type of metadata is based directly on the
contents of an asset. A popular example of this is full-text indices based on the text
of the documents. Inverted tree and document vectors are examples of this type of
metadata.
Content-descriptive metadata: This type of metadata describes the contents of an
asset without their direct utilization. It often involves use of knowledge or human
perception or cognition. An example is denoting the fragrance of an image
containing a flower. Another example of this type of metadata is textual annotations
describing the contents of an image.
Domain-independent metadata: This type of metadata captures information present
in the document independent of the application or subject domain of the information.
Examples of these are the C/C++ parse trees and HTML/SGML document type
definitions.
Domain-specific metadata: Metadata of this type are described in a manner specific
to the application or subject domain of information. Issues of vocabulary become
very important in this case as the terms have to be chosen in a domain-specific
manner. Examples of such metadata are relief, land cover from the GIS domain, and
area and population from the Census domain. In the case of structured data, the
database schema is an example of such metadata. Another interesting example is
domain-specific ontologies, terms from which may be used as vocabulary to
construct metadata specific to that domain.
Metadata may be precomputed (and possibly stored in a database) or they may be
computed when needed (at query-processing time), in which case they may be
represented by a computation, procedure, or method (e.g., an image processing
routine giving values for land-cover metadata of a satellite image, executed when
needed; Kashyap 1997).
Use of different types of metadata leads to different query or information access
options. For example, three forms of information access, as supported in the
VisualHarness system (Mudumbai 1997; Shah et al. 1997) are keyword-based
access, attribute-based access, and content-based access. Table 2 shows relationships
between various aspects of metadata for image data as exploited in the
VisualHarness system. Furthermore, information access may involve iterative and
customized (through use of different relative weights) use of one or more of these,
as shown in Figure 5. Other types of information access options are possible,
including standard, mediated, and immersive browsing (Grosky et al. 1998),
semantic associative search (Kiyoki et al. 1998), and mixed media (Chen et al.
1998). Understanding the role of metadata and its use is likely to be one of the most
enduring legacies of this generation of systems.
Some of the systems in this generation adapted the FDBS architecture to
federated information systems architecture, where the DBMSs as managers of
component information sources (databases) were replaced by a broader variety of
information systems, including simple access protocols to access data, a broad
variety of DBMSs (from network DBMSs to relational DBMSs to object-relational
and object-oriented DBMSs), DBMSs specialized to manage specific digital media
types (predominantly images), the Web itself as a manager of semi-structured data,
Changing Focus on Interoperability in Information Systems 15

and even expert systems. However, the mediator architectures (Wiederhold 1992)
were clearly the dominant ones, involving wrappers for encapsulating heterogeneous
information sources to provide more uniform interface to the rest of the world, and
mediators to provide a broad variety of value-added services (Wiederhold 1997).

Metadata Content Semantic Typical Primary Example for the

proc. (domain- format of access metadata of this type
used specific) metadata option from image data
info. used
Content- No No Attribute Attribute- Image height, width, size,
indep. values based date of creation, etc.
Content- Yes No Feature Content- Color, composition,
dep. values based texture, structure etc.
(domain-independent)
Content- No Yes Attribute Attribute- Color (based on human
based values based interpretation), height (of
the real-world object
depicted by the image),
fragrance, hybridizer of a
flower; model, category
etc. of an aircraft
(domain-specific)
Content- May be Either Text and Keyword- Descriptions of a flower,
descr. indices on based an x-ray, an aircraft
text (domain-specific);
general descriptions of
the image itself (domain-
independent)

Table 2: Image metadata classification and uses

During this generation, we saw increasing standardization or adoption of ad hoc

standards, resulting in significant progress towards achieving system, syntactic, and
structural interoperability. Acceptance of the Internet as a standard for
interconnections between the systems, and evolution of infrastructures and
middleware that support distributed computing (RMI, CORBA, and DCOM), and
database connectivity (e.g., ODBC, JDBC etc. for relational databases), have had
significant positive impact on achieving system interoperability. Syntactic
interoperability includes the ability to deal with formatting and data exchange as
supported by standards such as HTML for most current web-accessible documents,
Z39.50 for bibliographic data, and MPEG-1 for pixel-level representation of image
data. At the structural level, standardization for data modeling such as ANSI SQL,
and object modeling standards and methodologies such as ODMG and UML have
helped. Structural interoperability is further promoted by the use of IDL for
distributed objects. Structural and a limited form of semantic interoperability are
achieved by adoption of general purpose metadata standards, such as Dublin Core
(OCLC 1997), as well as metadata standards in various domains such as
bibliography (e.g., LCC and DDC; Beard and Smith 1998), space and astronomy,
16 Amit P. Sheth

and geographical and environmental (e.g., FGDC and UDK; Günther and Voisard
1998).
Among a large number of systems representing this generation, three classes of
systems stand out: systems focusing on information integration or uniform access to
heterogeneous repositories, systems providing more dynamic architecture or query
processing mechanisms to process a user service or information request on demand,
and systems that address domain-specific or semantic-level issues. A brief review of
these three classes of systems follows.
Changing Focus on Interoperability in Information Systems 17

Figure 4. Keyword-, attribute-, and content-based access in VisualHarness

A small representative set of systems that support access to heterogeneous and

distributed information sources includes TSIMMIS (http://www-
18 Amit P. Sheth

db.stanford.edu/tsimmis), Information Manifold (http://www.research.att.com/~levy

/imhome.html), GARLIC (http://www.almaden.ibm.com/cs/showtell/garlic),
InfoHarness/VisualHarness (http://lsdis.cs.uga.edu/infoharness), SIMS (http://
www.isi.edu/sims), HERMES (http://www.cs.umd.edu/projects/hermes), and
InfoSleuth (http://www.mcc.com:80/projects/infosleuth). SIMS, HERMES, and
InfoSleuth also share capabilities with the third class as they support domain
modeling or ontology, coupled with knowledge-based reasoning. WebSQL is an
example of the second class of systems. There are also a number of systems that
focus on specific media types or application domains, for example, QBISM (Arya et
al. 1994) and CoBase (http://burton.cs.ucla.edu/) for image data management, and
THETIS (Houstis et al. 1997) and Drew and Ying (1998) for environmental data and
geographical information systems, etc. For brevity, we only discuss some of these
systems.
TSIMMIS (Garcia-Molina et al. 1995) uses a mediator approach to combine
information from several sources containing textual and semi-structured data. Data
sources are encapsulated using wrappers or translators that logically convert the data
to a common information model by translating information requests and results to
this common model. The mediator layer above the wrappers is responsible for
routing queries to appropriate sources and for post-processing the results. An
important focus of the system is to generate wrappers and mediators automatically
for a set of specified rules. Thus, TSIMMIS provides a framework for users to
specify information integration, which may be done manually or in a semi-automated
manner.
The Information Manifold (Levy et al. 1995) is a system for retrieval and
organization of information from disparate (structured and unstructured) information
sources. The architecture of Information Manifold is based on a knowledge base
containing a rich domain model that enables describing the properties of the
information sources. The user can interact with the system by browsing the
information space (which includes both the knowledge base and the information
sources). The presence of descriptions of the information sources also enables the
user to pose high-level queries based on the content of the information sources. One
area of focus in the project has been to optimize the execution of a user query
expressed in a high-level language, which might potentially require access to and
combination of content from several information sources.
The InfoHarness system (Shklar et al. 1995) and its commercial counterpart
AdaptX/Harness (Sheth 1996) provide browsing as well as keyword- and attribute-
based access (involving typed attributes and logical operators) to a broad variety of
Web-accessible heterogeneous documents (e.g., unstructured text, semistructured
text such as AP news and emails, word processor and source code files, data
accessible through well-defined interfaces such as NNTP news group server), and
relational databases. Some of the key features are extensibility of metadata and
corresponding dynamically-created query interface with typed attributes, logical
remodeling of information space (such as browsing or searching source code files by
function signatures), and support for multiple, third-party indexing strategies. The
VisualHarness system (Mudumbai 1997; Shah and Sheth 1998) further extended it to
support images using third-party visual image retrieval engines, and customizable
access involving keyword-, attribute-, and content-based access.
Changing Focus on Interoperability in Information Systems 19

WebSQL (Mendelzon et al. 1997a) is a high-level declarative query language for

extracting information from the Web. WebSQL takes advantage of multiple index
servers without requiring users to know about them, and integrates full text with
topology-based queries. This enables definition of the content of domain-specific
text indexes. WebSQL is used to define logical views on the unstructured global
repository of Web-accessible documents. A level above Web SQL is the Web
Semantics Query Language (WSQL) (Mendelzon et al. 1997b). This system has
different layers of abstraction and provides mechanisms for describing the data that
are available, for discovering the existence of data relevant to a problem, and for
accessing discovered relevant data. WSQL has constructs for source discovery via
controlled Web navigation, source registration in domain-specific catalogs,
associative selection of sources from existing catalogs, and uniform access to data
stored in heterogeneous sources.
There are several systems that employ the metadata-based semantic view of the
world and employ an ontological layer above it. In the SIMS project (Arens et al.
1996) a model of the application domain is created using a knowledge representation
system to establish a fixed vocabulary describing objects in the domain, their
attributes, and relationships among them. For each information source a model is
constructed that indicates the data model used, query language, network location,
size estimates, etc., and describes the contents of its fields in relation to the domain
model. Queries to SIMS are written in the high-level uniform language of the
domain model. SIMS determines the relevant information sources by using the
knowledge encoded in the domain model and the models of the information sources.
These information sources are determined at run time based on their availability at
that time.
The HERMES (Adali and Subrahmanian 1994) system follows the mediator
architecture for semantically integrating different and possibly heterogeneous
information sources (including those containing visual data) and reasoning systems.
This integration is done using mediators that are very similar to the ones in the
TSIMMIS system described above. Mediators, in HERMES, are logical guidelines of
how information from different sources will be combined and integrated. In this
framework, external information sources are abstracted as domains which execute
certain functions with pre-specified input and output types. These domains are
accessed in mediators using a logic-based declarative language. The system also
provides a uniform environment for adding new external sources to existing
mediators.
The InfoSleuth system (Bayardo et al. 1997) views an information source at the
level of its relevant semantic concepts, thus preserving the autonomy of its data.
Information requests to InfoSleuth are specified generically, independent of the
structure, location, or even existence of the requested information. InfoSleuth filters
these requests, specified at the semantic level, flexibly matching them to the
information resources that are relevant at the time the request is processed. The
InfoSleuth approach is to specify a common ontology for a domain, and local
mappings from individual database schemas to the common ontology. These
mappings can be thought of as views of the data that simplify query specification for
selecting information. Given an appropriate set of mappings for a particular
knowledge discovery task, the InfoSleuth system provides query support for
20 Amit P. Sheth

selecting relevant information. It also pre-processes and transforms the underlying

database data into records whose attributes consist of concepts from the ontology.
Early emphasis in the InfoSleuth system was on harnessing structured databases and
support for text data is also reported.

4. Third generation

One thing that has not changed for the third generation is that we are once again
faced with more distribution, more autonomy, and more heterogeneity among the
accessible information, information sources, and users. With the progress in global
interconnectivity, we now need to deal with more heterogeneous information
consisting not only of a broader variety of digital data, but also operations and
computations (such as simulations) that can create new data and information. The
scale of the problem has changed from a few databases to millions of information
resources, and the new resources are added independently to the accessible set of
resources, as other resources change rapidly or disappear. Currently favorite
strategies that depend on keyword-based access or involve only representational or
structural components of data are usually found to provide a poor quality of result,
and their lack of precision leads to increasing information overload. We fully expect
increasing standardization and interoperability at system, syntactic, and structural
levels to address many issues—for example, see Paepcke et al. (1998) for relevant
work in the domain of digital libraries. However, the key challenges to be faced are
at the semantic level, where people would increasingly expect the information
systems to help them not at the data level, but at the information, and increasingly
knowledge levels.
Even a casual user of the Web is aware of the rapid increase in the amount and
diversity of information available online. However, what is creating an even bigger
challenge is the increased expectations of the user in terms of understanding of the
context of the user’s information need, increasing availability of semantically rich
visual and new media, and a corresponding need to support semantic-level
interoperability. The problem of information overload has turned the challenge of
“So far (schematically) yet so near (semantically)” (Sheth and Kashyap 1993) faced
by the previous generations into “So near (syntactically and structurally) yet so far
(semantically)”.
Although there are several uses and interpretations of semantics in information
systems, our view is that future information systems will need to support a more
general notion that involves relating the content and representation of information
resources to entities and concepts in the real world (Beech 1997; Meersman 1997;
Sheth 1997). That is, the limited forms of operational and axiomatic semantics of a
particular representational or language framework are not sufficient (see Paepcke et
al. 1998 for a relevant discussion on syntax and some types of semantics). Semantic
interoperability will then support high-level (hence easier to use), context-sensitive
information requests over heterogeneous information resources, hiding system,
syntax, and structural heterogeneity. In essence, we need an approach that reduces
the problem of knowing the contents and structure of many information resources to
the problem of knowing the contents of easily-understood, domain-specific
Changing Focus on Interoperability in Information Systems 21

ontologies, which a user familiar with the domain is likely to know or understand
easily.
Foundational research leading to building the third generation of information
systems has been carried out in several umbrella projects and initiatives, including
Knowledge Sharing Effort (http://www-ksl.stanford.edu/knowledge-sharing),
Intelligent Integration of Information (http://mole.dc.isx.com/I3), and the Digital
Library Initiative (http://www.cise.nsf.gov/iis/dli_home.html). Systems belonging to
the third class of the second generation have also made contributions that the third
generation systems can build on. Increasing standardization at different levels of
information systems architecture for corresponding type of interoperability also plays
an important role. Some of the examples are as follows.
• System: IIOP for interactions between distributed objects and components,
KQML for interaction between agents;
• Syntactic: XML for all forms of Web-accessible data;
• Structural: RDF for general purpose description of information sources, various
object models for web-based information exchange (Manola 1998), MPEG-4 for
structural or object-level description video, MHEF-5 for multimedia and
hypermedia, KIF for knowledge representation, OKBC for distributed
knowledge bases;
• Semantic: MPEG-7 (still in progress) with likely support for limited forms of
semantics with identification of context, objectives requirements, and
applications.
We now focus our attention on a discussion of possible enablers of semantic
interoperability. In particular, we identify three enablers and capabilities:
Terminology (and language) transparency: This will allow a user to choose an
ontology of his or her choice (e.g., one based on LCC for querying bibliographic data
or FGDC for geospatial data), while allowing the information source to subscribe to
a related but different ontology (e.g., an ontology based on DDC or UDK,
repsectively. The latter recognizes some overlap between geospatial data sets and
environmental data sets, and their respective modeling).
Context-sensitive information processing: The information system will recognize
or understand the context of an information need and use it to limit information
overload, both by formulating more precise queries used for searching information
sources and by filtering and transforming the information before presenting it to the
user.
Semantic correlation: This will allow the representation of semantically-related
information regardless of distribution and heterogeneity (including various forms of
media) by the user or the third party, and their use for obtaining all forms of relevant
information anywhere.
Three key components of a possible solution are metadata (especially domain-
specific and content-based metadata), contexts, and ontologies (Kashyap and Sheth
1998). We briefly discuss their role in developing semantic interoperability
solutions. One key aspect of the third generation (operation or process
interoperation) will not be discussed for brevity.
22 Amit P. Sheth

4.1. Ontologies and terminology transparency

An ontology can be defined as a specific vocabulary and relationships used to

describe certain aspects of reality, and a set of explicit assumptions regarding the
intended meaning of the vocabulary of words (Gruber 1991; Guarino 1998). Among
various other classification schemes and structures, including keywords, thesauri,
and taxonomies, ontologies are often viewed as allowing more complete and precise
domain models (Huhns and Singh 1997). Support and use of multiple,
independently-developed ontologies is important for developing scalable information
systems with multiple information producers and consumers (e.g., Arens et al. 1996;
Dao and Perry 1996; Genesereth and King 1995; Kashyap and Sheth 1998; Khang
and McLeod 1998 for need and use of multiple ontologies). One challenging issue in
supporting semantic interoperability is how to allow both users and providers to
subscribe to existing ontologies of their choice or create a new one (Kashyap and
Sheth 1998). Processing an information request represented in terms of one ontology
in an environment with information resources that subscribe to different (but related
and relevant) ontologies may involve using inter-ontological relationships, such as
synonym, hypernym, homonym, and other possibly domain-specific relationships.
This work also requires understanding of and containing loss of information in multi-
ontology query processing (Mena et al. 1998). One early example of research along
these lines is the OBSERVER (sub)system (http://siul02.si.ehu.es/~jirgbdat
/OBSERVER), which is a component of the InfoQuilt system
(http://lsdis.cs.uga.edu/infoquilt).

4.2. Context

In characterizing the similarity between objects based on the semantics associated

with them we have to consider the real-world semantics (RWS) of an object. It is not
possible to completely define what an object denotes or means in the model world.
We propose the context of an object as the primary vehicle to capture the RWS of the
object. Understanding of the context of the information request can help the system
to distinguish between whether the term cricket refers to an insect or a sports game.
Adapting from research in AI and Knowledge-Based systems (e.g., Shoham
1991), linguistics and other fields, modeling and representing context can lead to
several benefits in dealing with information overload in a global information
infrastructure (GII; see Kashyap and Sheth 1998 for more details):
• Economy of representation: In a manner akin to database views, contexts can act
as a focusing mechanism when accessing the component databases or
information sources on the GII.
• Economy of reasoning: Instead of reasoning with the information present in the
database as a whole, reasoning can be performed with the context associated
with an information source.
• Managing inconsistent information: In the GII, where information sources are
designed and developed independently, it is not uncommon to have information
in one source be inconsistent with information in another. As long as
Changing Focus on Interoperability in Information Systems 23

information is consistent within the context of the query of the user,

inconsistency in information from different databases may be allowed.
• Flexible semantics: An important consequence of associating abstractions or
mappings with context is that the same two objects can be related to each other
differently in two different contexts. Two objects might be semantically closer
to each other in one context as compared to the other.
There are several proposals for representing context. We believe that an effective
approach needs to bring together metadata, user profiles, information modeling
abstractions, and ontologies, as well as to allow their dynamic construction to model
application domain and user needs. Besides their modeling and representation, a key
challenge includes the ability to reason about or compare contexts (e.g., Kashyap and
Sheth 1996; Lee et al. 1996; Ouksel and Naiman 1994). While there are many
representations and associated reasoning techniques, practical application of context
in GII is expected to be a key research challenge for achieving semantic
interoperability in information systems.

4.3. Information correlations

One of the key applications of semantics in GII is to represent or specify information

requests and semantic level information correlations regardless of the media (and
other heterogeneity) and locations of information sources. These can involve queries
over heterogeneous media assets represented at a higher level of abstraction in
media-independent manner, using metadata and ontologies.
Two approaches to representing information correlations between independently-
managed networked resources are Metadata Reference Links (MREFs; Shah and
Sheth 1998; Sheth and Kashyap 1996) and Distributed Active Relationships (DARs;
Daniel et al. 1998). They provide an initial step in specifying information correlation
between heterogeneous digital media. Specifically, MREFs allow subscription to one
or more ontologies in their specification, and the meta-information used in
specifying an MREF is mapped to views involving keyword-based, attribute-based,
and content-based specifications involving various types of metadata of
heterogeneous digital media. Specification and processing based on information
correlations can be easily integrated with the Web technology. For example, MREF
could be used anywhere a hypermedia link (HREF) is used, and its specification and
processing can be supported using an RDF and XML-based infrastructure. However,
many challenges remain in extending the current proposals to include non-standard
resources such as datasets and procedures, integrating information correlation
representation and processing with context and context mediation, and processing
them efficiently in a very large information space.

4.4. Information-brokering architecture

It is hard to predict the architecture of the third generation of information systems.

One proposed architecture, information brokering (Kashyap 1997; Kashyap and
Sheth 1994), adapts and extends the concepts of (1) federated environments
(Heimbigner and McLeod 1985; Sheth and Larson 1990) in which resources,
24 Amit P. Sheth

metadata, and ontologies are created, administered, and enhanced independently; and
(2) mediator architectures (Wiederhold 1992) which involve decoupling information
creators and providers from information users and better semantic-level services and
interoperability. However, we believe that the key to this generation of systems is
their support for semantic interoperability, through exploitation of various forms of
metadata, multiple ontologies, and contexts. Furthermore, we believe that before
very general architectures that can support various domains can be developed,
support for semantic interoperability demands that we focus on a specific domain
first, such as GIS (Goodchild et al. 1997), and then extend what we learn to general-
purpose and multi-domain environments. Figure 6 shows a schematic of a system for
supporting semantic interoperability as described above in a geographical domain. It
is too early to give representative examples of the third generation, but a few early
efforts are described by Wiederhold (1996), and Papozoglou and Schlageter (1998);
and see InfoQuilt (http://lsdis.cs.uga.edu/infoquilt).

Figure 5. An architecture to support geographic information brokering

In closing, we believe that semantic interoperability is the key to progress towards

a vision of Infocosm, a society whose members will have information anywhere, any
time, and in many forms, for knowledge creation and use, effective decision-making,
better learning, and more fun.

Acknowledgements

Members of the InfoQuilt project, especially Vipul Kashyap, Kshitij Shah, Clemens
Bertram, and Krishnan Parsuraman, have contributed to some of the ideas presented
in this paper.
Changing Focus on Interoperability in Information Systems 25

References

Adali S, Subrahmanian V S 1994 Amalgamating knowledge bases, II: Distributed mediators.

International Journal of Intelligent and Cooperative Information Systems 3(4): 349–383
Arens Y, Knoblock C A, Shen W 1996 Query reformulation for dynamic information
integration. In Wiederhold G (ed) Intelligent Integration of Information. Kluwer Academic
Publishers: 11-42
Arya M, Cody W, Faloutsos C, Richardson J, Toga A 1994 QBISM: extending a DBMS to
support 3D medical images. Proceedings of the 10th International IEEE Conference on
Data Engineering, Houston, Texas: 314-325
Batini C, Lenzerini M, Navathe S 1986 A comparative analysis of methodologies for database
schema integration. ACM Computing Surveys 18(4): 323–64
Bayardo R, Bohrer W, Brice R, Cichocki A, Fowler G, Helal A, Kashyap V, Ksiezyk T,
Martin G, Nodine M, Rashid M, Rusinkiewicz M, Shea R, Unnikrishnan C, Unruh A,
Woelk D 1997 Semantic integration of information in open and dynamic environments.
Proceedings of the 1997 ACM International Conference on the Management of Data
(SIGMOD), Tucson: 195-206.
Beard K, Smith T 1998 A framework for meta-information in digital libraries. In Sheth A,
Klas W (eds) Multimedia Data Management: Using Metadata to Integrate and Apply
Digital Media. McGraw Hill: 341-365
Beech D 1997 Data semantics on the information superhighway. In Meersman R, Mark L
(eds) Database Application Semantics. Chapman and Hall
Boll S, Klas W, Sheth A 1998 Overview on using metadata to manage multimedia data. In
Sheth A, Klas W (eds) Multimedia Data Management: Using Metadata to Integrate and
Apply Digital Media. McGraw-Hill Publishers: 1–23
Chen F, Hearst M, Kimber D, Kupiec J, Pedersen J, Wilcox L 1998 Metadata for mixed-
media access. In Sheth A, Klas W (eds) Multimedia Data Management: Using Metadata to
Integrate and Apply Digital Media. McGraw Hill: 319-340
Daniel R, Lagoze C, Payette S 1998 A metadata architecture for digital libraries. Proceedings
of the Forum on Research and Technology Advances in Digital Libraries (ADL'98), Santa
Barbara: 276-288
Dao S, Perry B 1996 Information mediation in cyberspace: scalable methods for declarative
information networks. In Wiederhold G (ed) Intelligent Integration of Information. Kluwer
Academic Publishers: 43-62
Drew P, Ying J 1998 Metadata management for geographic information discovery and
exchange. In Sheth A, Klas W (eds) 1998 Multimedia Data Management: Using Metadata
to Integrate and Apply Digital Media. McGraw Hill: 89-121
Drew P, King R, McLeod D, Rusinkiewicz M, Silberschatz A 1993 Report of the workshop on
semantic heterogeneity and interoperation in multidatabase systems. SIGMOD Record
22(3): 47–56
Elmagarmid A (ed) 1992 Database Transaction Models for Advanced Applications. Morgan-
Kaufmann
Elmagarmid A, Pu C (eds) 1990 Heterogeneous databases. Special Issue. ACM Computing
Surveys 22 (3)
Elmagarmid A, Sheth A, Rusinkiewicz M (eds) 1998 Heterogeneous Distributed Databases.
Morgan Kaufmann Publishers
Garcia-Molina H, Hammer J, Ireland K, Papakonstantinou Y, Ullman J, Widom J 1995
Integrating and accessing heterogeneous information sources in TSIMMIS. Proceedings of
the AAAI Symposium on Information Gathering, Stanford, California: 61-64
Genesereth M, King R (eds) 1995 Reference Architecture, Intelligent Integration of
Information. Stanford University and University of Colorado. http://logic.stanford.edu
/achitecture/reference.html
26 Amit P. Sheth

Georgakopoulos D, Rusinkiewicz M 1994 Using tickets to enforce the serializability of

multidatabase transactions. IEEE Transactions On Knowledge and Data Engineering 6(1):
166–180
Goodchild M, Egenhofer M, Fegeas R 1997 Interoperating GISs: Report of a Specialist
Meeting Held under the Auspices of the Varenius Project, Panel on Computational
Implementations of Geographic Concepts. National Center for Geographic Information and
Analysis, Santa Barbara
Grosky W, Fotouhi F, Jiang Z 1998 Using metadata for the intelligent browsing of structured
media objects. In Sheth A, Klas W (eds) Multimedia Data Management: Using Metadata
to Integrate and Apply Digital Media. McGraw Hill: 123-148
Gruber T 1991 The role of a common ontology in achieving sharable, reusable knowledge
bases. Proceedings of the 2nd International Conference on Principles of Knowledge
Representation and Reasoning, Cambridge: 601-602
Guarino N 1998 Formal ontology and information systems. Proceedings of the 1st
International Conference on Formal Ontology in Information Systems [FOIS’98], Torino:
3-15
Günther O, Voisard A 1998 Metadata in geographic and environmental data management. In
Sheth A, Klas W (eds) Multimedia Data Management: Using Metadata to Integrate and
Apply Digital Media. McGraw Hill: 57-87
Gupta A, Jain R 1997 Visual information retrieval. Communications of the ACM 40(5): 70–79
Heimbigner D, McLeod D 1985 A federated architecture for information management. ACM
Transactions on Office Information Systems 3(3): 253–278
Houstis C, Nikolaou C, Marazakis M, Patrikalakis N, Sairamesh J, Thomasic A 1997
THETIS: Design of a data repositories collection and data visualization system for coastal
zone management of the Mediterranean Sea. D-Lib Magazine, The Magazine for Digital
Library Research, November. http://www.dlib.org/dlib/november97/thetis/11thetis.html
Hsiao D, Neuhold E, Sacks-Davis R (eds) 1993 Interoperable Database Systems (DS-5.,
North-Holland
Huhns M, Singh M 1997 Ontologies for agents. Internet Computing 1(6): 81-83
Kambayashi Y, Rusinkiewicz M, Sheth A (eds) 1991 Proceedings of the RIDE-IMS'91:
International Workshop on Interoperability in Multidatabase Systems. IEEE Computer
Society
Kashyap V 1997 Information Brokering over Heterogeneous Digital Data: A Metadata Based
Approach. PhD Thesis, Rutgers University
Kashyap V, Sheth A 1994 Semantics based information brokering. Proceedings of the 3rd
International Conference on Information and Knowledge Systems: 363-370
Kashyap V, Sheth A 1996 Schematic and semantic similarities between database objects: a
context-based approach. The Very Large Databases Journal 5(4): 276-304
Kashyap V, Sheth A 1998 Semantic heterogeneity in global information systems: the role of
metadata, context and ontologies. In Papazoglou M, Schlageter G (eds) Cooperative
Information Systems: Current Trends and Directions. Academic Press: 139-178
Kashyap V, Shah K, Sheth A 1995 Metadata for building the MultiMedia Patch Quilt. In
Jajodia S, Subrahmanian V S (eds) Multimedia Database Systems: Issues and Research
Directions. Springer-Verlag: 297-319
Khang J, McLeod D 1998 Dynamic classificational ontologies: mediation of information
sharing in cooperative federated database systems. In Papazoglou M, Schlageter G (eds)
Cooperative Information Systems: Current Trends and Directions. Academic Press: 179-
203
Kim W (ed) 1995 Modern Database Systems: The Object Model, Interoperability and Beyond.
Addison Wesley
Kim W, Choi I, Gala S, Scheevel M 1993 On resolving schematic heterogeneity in
multidatabase systems. Distributed Parallel Databases International Journal 1: 251-279
Changing Focus on Interoperability in Information Systems 27

Kiyoki Y, Kitagawa T, Hayama T 1998 A metadatabase system for semantic image search by
a mathemetical model of meaning. In Sheth A, Klas W (eds) Multimedia Data
Management: Using Metadata to Integrate and Apply Digital Media. McGraw Hill: 191-
222
Lagoze C, Lynch C, Daniel R 1996 The Warwick Framework: A container architecture for
aggregating sets of metadata. Technical Report TR96-1593. Cornell University,
Department of Computer Science. http://cs-tr.cs.cornell.edu:80/Dienst/UI/2.0/
Describe/ncstrl.cornell/ TR96-1593
Lee J, Madnick S, Siegel M 1996 Conceptualizing semantic interoperability: a perspective
from the knowledge level. International Journal of Cooperative Information Systems 5(4):
367–393
Levy A Y, Srivastava D, Kirk T 1995 Data model and query evaluation in global information
systems. Intelligent Information Systems 5(2): 121-143
Litwin W, Boudenant J, Esculier C, Ferrier A, Glorieux A, La Chimia, J, Kabbaj K,
Moulinoux C, Rolin P, Stangret C 1982 SIRIUS: systems for distributed data management.
In Schneider H-J (ed) Distributed Data Bases. North-Holland, Netherlands: 311–66
Manola F 1998 Towards a Web Object Model. Object Services and Consulting, Inc.
http://www.objs.com/OSA/wom.htm
Meersman R 1997 An essay on the role and evolution of data(base) semantics. In Meersman
R, Mark L (eds) Database Application Semantics. Chapman and Hall
Meersman R, Mark L (eds) 1997 Database Application Semantics. Chapman and Hall
Mena E, Kashyap V, Illarramendi A, Sheth A 1998 Domain specific ontologies for semantic
information brokering on the global information infrastructure. Proceedings, International
Conference on Formal Ontology in Information Systems (FOIS'98), Torino: 269-283
Mendelzon A, Mihaila G, Milo T 1997a Querying the World Wide Web. Journal of Digital
Libraries 1(1): 68–88
Mendelzon A, Mihaila G, Raschid L, Tomasic A 1997b Locating and accessing heterogeneous
data sources. Proceedings of CASCON'97
Mudumbai S 1997 ZEBRA: Customizable, Extensible Metadata-based Access to Federated
Image Repositories. M.S. Thesis, Department of Computer Science, University of Georgia
Online Computer Library Center, Inc 1997 Dublin Core Metadata Element Set: Reference
Description. Office of Research and Special Projects, Dublin, Ohio. http://
www.oclc.org:5046/research/dublin_core/
Ouksel A, Naiman C 1994 Coordinating context building in heterogeneous information
systems. Journal of Intelligent Information Systems 3 (2): 151-183
Paepcke A, Chang C, Garcia-Molina H, Winograd T 1998 Interoperability for digital libraries
worldwide. Communications of the ACM 41(4): 33-43
Papazoglou M, Schlageter G (eds) 1998 Cooperative Information Systems: Current Trends
and Directions. Academic Press
Ram S (ed) 1991 Heterogeneous distributed database systems. Special Issue. IEEE Computer
24(12)
Schek H–J, Sheth A, Czjedo B (eds) 1993 Proceedings of the RIDE-IMS'93: International
Workshop on Interoperability in Multidatabase Systems. IEEE Computer Society
Shah K, Sheth A 1998 Logical information modeling of Web-accessible heterogeneous digital
assets. Proceedings of the Forum on Research and Technology Advances in Digital
Libraries (ADL'98), Santa Barbara: 266-275
Shah K, Sheth A, Mudumbai S 1997 Black box approach to image feature manipulation used
by visual information retrieval engines. Second IEEE Metadata Conference
Sheth A 1987 Heterogeneous Distributed Databases: Issues in Integration. Tutorial Notes,
3rd International Conference on Data Engineering
Sheth A (ed) 1991 Semantic issues in multidatabase systems. Special Issue. SIGMOD Record
28 Amit P. Sheth

Sheth A 1995 Multidatabase Interoperation: Perspective of Researchers and Practitioners.

Tutorial Notes, 11th International Conference on Data Engineering, Taiwan
Sheth A 1996 Bellcore's ADAPT/XHarness system for managing information on Internet and
intranets. Proceedings of the 22nd International Conference on Very Large Data Bases,
Bombay, India: 585
Sheth A 1997 Data semantics: What, where, and how? In Meersman R, Mark L (eds)
Database Application Semantics. Chapman and Hall: 601-610
Sheth A, Kashyap V 1993 So far (schematically) yet so near (semantically). In Hsiao D,
Neuhold E, Sacks-Davis R (eds) 1993 Interoperable Database Systems (IFIP Transaction
A-25, Proceedings of DS-5) North-Holland: 283-312
Sheth A, Kashyap V 1996 Media-independent correlation of information: what? how?
Proceedings of First IEEE Metadata Conference
Sheth A, Klas W (eds) 1998 Multimedia Data Management: Using Metadata to Integrate and
Apply Digital Media. McGraw Hill
Sheth A, Larson 1990 Federated database systems for managing distributed, heterogeneous,
and autonomous databases. ACM Computing Surveys 22(3): 183–236
Shklar L, Sheth A, Kashyap V, Shah K 1995 InfoHarness: Use of automatically generated
metadata for search and retrieval of heterogeneous information. Proceedings of CAiSE-95:
217-230
Shoham Y 1991 Varieties of context. In Lifschitz V (ed) Artificial Intelligence and
Mathematical Theory of Computation: Papers in Honor of John McCarthy. Academic
Press, Boston: 393–408
Wiederhold G 1992 Mediators in the architecture of future information systems. IEEE
Computer 25(3): 38–49
Wiederhold G (ed) 1996 Intelligent Integration of Information. Kluwer Academic Publishers
Wiederhold G 1997 Value-added mediation in large-scale information systems. In Meersman
R, Mark L (eds) Database Application Semantics. Chapman and Hall