1 Introduction
The Large Scale Distributed Information Systems Lab (LSDIS) at the
Enabling Infocosm Theme. The merging of computers and communications with significant advances
in networking infrastructure has made millions of information sources with wide
varieties of information accessible. This capability gives us a vision of an
information rich society "infocosm", where we expect to have any information any where we want in
(m)any form(s) for effective decision making and knowledge-centric activities,
improved productivity, and fun. Emphasis of our research is on
semantics (meaning and use) of information and semantic interoperability
supported by an information brokering architecture. This is closely tied to the recent interest
in Semantic Web, which is presented as “the Web of data (and connections) with
meaning in the sense that a computer program can learn enough about what the
data means to process it (Tim Berners-Lee, Weaving the Web, 1999). Our emphasis is on organizing and utilizing Web-based as well as enterprise content
semantically, rather than though syntactic and structural methods.
Research in enabling Infocosm through semantic information brokering is
anchored by the InfoQuilt, project, with InfoHarness/VisualHarness, ADEPT, and
Video Anywhere as associated projects (Section 2).
Processes for Networked
Organizations. The interoperability of information
systems, applications and users within and across enterprises has important
implications on the competitiveness of organizations in the global economy.
Three important components of a comprehensive approach to enterprise
integration are: integration and interoperability of applications and
information systems (as in Enterprise Application Integration), coordination of
activities through automated process and workflow management, and collaboration
between humans as well as effective interactions between humans, applications
and information systems. Primary project in this theme has been METEOR, with
CaTCH as an associated project, and research as well as applications in
Bioinformatics (Section 3).
An important aspect of LSDIS’s activities is that research,
technology development and prototyping efforts have occurred in close
collaboration with industry and government partners. We strive to achieve
impact of our research not only through publications and invited talks, but
also through industry trials, real-world applications, technology transfers and
commercialization. Our significant
success in this respect is summarized in Section 4.
2 Enabling Infocosm: InfoQuilt and
related projects
InfoQuilt’s objective
is to enable Infocosm, where we can support effective decision making and
improve productivity through a variety of semantic techniques to organize and
utilize all forms of data and information, on the Web and within an enterprise.
The InfoQuilt system uses a multi-agent information brokering architecture to
support the following capabilities:
n Access, analyze and interoperate with
heterogeneous, static or dynamic (pull and push) content (pages,
documents, sites, repositories, databases) using both wrapper and information
extraction technologies
n Semi-Automatically or Automatically create semantic metadata
(domain-specific or contextually relevant metadata), as well as the syntactic
metadata
n Use and support of multiple (possibly preexisting) ontologies and domain
modeling with complex inter-ontology relationships, domain rules and functional
dependencies
n User defined functions (esp. for fuzzy/approximate matching),
simulation; also used in post processing result analysis
n Information request processing utilizing domain semantics and resource
characteristics (local completeness, data characteristics, binding patterns)
At the conceptual level, we have been interested in
representation of context and semantic proximity [Kashyap and
Sheth 1996], use of semantic metadata and semantic information brokering
architecture [Kashyap and Sheth 2000], and the computation of information
loss in multi-ontology query processing [Mena et al 2000]. Our recent interest is in supporting knowledge discovery
from heterogeneous, autonomous sources of information, and supporting learning
through what-if analysis of empirically defined relationships between
data. To this end, we have introduced
the computing paradigm of ISCape (Information Scape) which allows users to query
and analyze the data available from a diverse autonomous sources, gain better
understanding of the domains and their interactions as well as discover and
study relationships. A
sample of IScape used as an Information Request is shown in Figure 1.
Use of IScape for
knowledge discovery can be exemplified using the following investigation
[Thacker et al 2001]:
“Do Nuclear Tests
result in Earthquakes?” Here a researcher has a hypothesis that this is the
case, and wants to prove or disprove it by studying independently collected
data made available at various Web-accessible information sources. The knowledge discovery process involves use
of ontologies related to Nuclear Tests and Earthquakes, and complex
relationships such as “causes” with spatial and temporal parameters.
Several projects are precursors to the InfoQuilt project.
InfoHarness/VisualHarness
The InfoHarness
system (1994-1997) [Shah and Sheth 1999] emphasized use of metadata in search
and retrieval of heterogeneous information in intranet/Internet environments.
The basic InfoHarness system, provided rapid access to huge amounts of
heterogeneous information on the Web and corporate intranets, without
reformatting, restructuring, or relocating the data. It also supported logical
restructuring of the information space, support for multiple third-party
indexing engines and many other features. Research in the joint Bellcore-LSDIS
project addressed several new research issues, including: keyword as well
attribute-based access to document collections, access to multiple autonomous
repositories with independent and heterogeneous indices for scaleable search
with associated intelligent merging of results, access to the same repository
with multiple indices to improve quality of result, and support for remote
servers using CORBA.
Figure 1: IScape for Information Correlation and
Request
VisualHarness system
(1997-1998) was the InfoHarness system enhanced to provide customizable and
extensible search of federated image repositories, in addition to text and
semi-structured data, and structured (relational) databases. Through the
comprehensive use of metadata of various types, it supported keyword,
attribute-based and content-based search over images. Also, different Visual
Information Retrieval (VIR) engines and third party indexing technologies could
be hooked into the system. A key idea was to form a combination of these
different access strategies to achieve better quality results. One interesting
technique developed was the black box approach for extracting the required information
from the VIR engine in the form we could access (distance metrics), without
knowing the internals of the VIR engine. We have shown the validity of this
black box approach by comparing the results obtained with this approach against
the results obtained from a VIR engine [Sheth et al 1999].
VideoAnywhere
This
industry-funded project researched the issues related to management of video in
embedded (e.g., cable set top) system running a Java Virtual Machine. It
targeted interactive TV and video application market-place. It involved
research in understanding metadata and consumer profile, caching, crawling
(using agent technology), etc. We also prototyped a Web-based search engine for
streaming audio and video on the Web, premium TV programming, and local video
(such as DVDs).
ADEPTUGA
UGA´s work on ADEPT - Alexandria
Digital Earth Prototype (1999-2000) involved: (a) specifying metadata for
geospatial information; (b) prototyping
IScape to support Digital Earth metaphor, incorporating complex and inter-ontological
relationships like “affects” and user defined functions including simulations,
and (c) development of a graphical knowledge management toolkit for easy
creation and deployment of IScapes, Ontologies and Relationships.
Looking ahead, our research
in multi-agent semantic information brokering is targeted at semantic-level solutions to serve a variety
of information stakeholders (participants), support learning and
decision-making in a more comprehensive manner, ease human – information system
interaction through semantics (such as better understanding of user’s context
and information need), and develop next generation of content management
regardless of source, format, media, representation and modeling of content.
3.
Workflow management is the automated coordination,
control and communication of work as is required to satisfy organizational
processes. A Workflow Management System (WFMS) is a set of tools providing
support for the necessary services of workflow process creation, workflow
enactment, and administration and monitoring of workflow processes. The METEOR system addresses the challenges of demanding
multi-organization processes and focuses on coordination of user and automated
tasks in real-world multi-enterprise heterogeneous computing environments.
METEOR [Managing End-To-End OpeRations]
The METEOR system consists of a suite of four
components (Figure 1):
Figure 1.
METEOR System Architecture
·
Builder for comprehensive, fully graphical
process modeling
·
Repository for workflow component browsing, reuse and
sharing
·
A choice of interoperating Enactment
Services:
- WebWork a fully distributed, Web based service [Miller et al 1998]
- ORBWork for fully distributed, CORBA based service [Kochut et al
1999] with recent support for Java/RMI distributed infrastructure instead of
CORBA
·
Manager for monitoring/administration
A unique aspect of the METEOR system is
automatic code generation for a heterogeneous, distributed environment from
graphical design/ specification. This can lead to 60 to 90% saving in
enterprise process application development and deployment, with corresponding
savings in costs and time.
METEOR has continually evolved as we
studied the requirements of healthcare, defense telecommunications and
e-commerce applications with our industry partners. Recent focus has been on supporting the
following capabilities:
High Security: Naval Research Lab (NRL) and LSDIS Lab worked together to provide security
within and across security domains (authentication, authorization, access
control, encryption and non-repudiation). Multilevel Security (MLS) features
are now part of the METEOR/NRL Workflow Builder and the ORBWork enactment
service.
High
Reliability and Survivability: METEOR’s
ingredients for achieving high reliability and survivability for workflows are
exception handling, recovery, and adaptability. Recent research in exception
handling has involved support for cross-organizational processes using case-based
reasoning, knowledge sharing, coordination, and intelligent problem solving.
Scalability: Once a workflow has been developed, it needs to be deployed, possibly
over a large geographical region and including multiple organizations.
ORBWork’s fully distributed architecture and implementation, utilizing CORBA,
Web and/or Java technologies, is highly scalable. ORBWork has configuration
specifications that make it easy to deploy workflow applications, as well as
move or relocate tasks at run time for performance improvement and adaptation.
Comprehensive
Modeling: We have investigated the relationships of workflow
modeling with database and simulation modeling [Miller, Sheth, Kochut 1999].
Rapid Design,
Development & Deployment: We have
developed a comprehensive repository (based on XML and object-relational
database technology) to store metadata about workflow designs, organizations,
informational resources (e.g., application databases) and computational
resources (CORBA servers, Java Servlets, or EJB components). The repository
enhances reusability and facilitates rapid incorporation of resources into
workflows, as well as supports adaptability.
Adaptability: Today’s military/commercial environments may rapidly undergo major
changes. Workflows must adapt to these
changes. The built-in adaptability of ORBWork combined with an easy-to-use
graphical designer, and a comprehensive repository make rapid adaptation
possible. In addition, monitoring tools as well as advanced exception handling
mechanisms are used to help indicating when adaptation may be necessary.
Inter-organizational
Workflows: The problem of multi-workflow enactment has many
facets, ranging from technological issues about how to integrate different
workflow management systems of different vendors on different platforms to the
purely conceptual issues of specifying how the interaction should occur, i.e.,
semantic interoperability. The enactment of different workflows, which can be
autonomously and separately designed, may cause problems such as deadlocks,
starvation, live-locks, or failure to terminate in the desired final state. The
available interoperability specifications, such as SWAP and JFlow help in
providing multi-workflows; but they are still far from meeting interoperability
needs of workflow systems.
Our current research directions involve
development of a multi-agent framework to support advanced capabilities for
inter-organizational workflows and Web services. Some current issues we are
investigating include Quality of Service specification, and survivability
techniques for federated Web services besides use of workflow techniques for
Web service integration and management.
CaTCH
While focus of METEOR
has been on coordination, collaboration is also a very important part of
enterprise processes, and was investigated in the CaTCH (Collaborative
Teleconsulting for Healthcare) project. The CaTCH project involved integration
for several rapidly progressing technologies to support high bandwidth
collaboration through access to diverse variety of information over a variety
of communications alternatives. CaTCH I supported real-time collaboration by
integrating multimedia patient data on Intranet, medical reference data on
Internet, LAN/POTS/ISDN-based Video+Data Conferencing, and WWW/Java programming
to set up remote environment and context sensitive collaboration. CaTCH II
supported asynchronous (store-and-forward) collaboration using a Web-object
model to integrate a variety of information such as streaming video of medical
information, video or audio mail, patient data, etc. and making it available on
demand through Web.
Bio-informatics
Research and Applications
Bio-informatics research constitutes an
important area in which LSDIS contribute extensively, especially in management
of genomics data and automating complex workflows in genome labs. In a joint
project with UGA Genetics Department, LSDIS is working to develop IntelliGEN, a
comprehensive system for genomic data and process management. An important goal
of this project is to discover interactions and roles of proteins in an
organism (i.e. Neurospora crassa).
Manual handling of
plate tracking, data collection, and computation of the protein-protein
interaction map processing of the planned 500 bait library plates is extremely
labor intensive and error-prone. In addition, most of the genomics and
bioinformatics tasks involved in the proposed project are very long and
complex. To address this problem, we are building upon our extensive work on an
extensible data and process support management system for handling the
experiments in protein-protein interaction mapping.
We have developed an automated workflow system called GENEFLOW [Hall et
al 2001] as well as the Fungal Genome Database (FGDB) in order to build a
laboratory information system for managing distributed high throughput
sequencing. In addition, we have created
graphical tools to visualize the mapping and sequencing data. These existing
graphical database tools support XML messaging to exchange genomic information
with other databases and applications. While earlier workflow systems have been
used to automate lab experiments, we believe that current advances in adaptive
workflow technologies can improve dramatically the quality of experiments by
optimizing laboratory workflows.
In the
near term, the core objective of the proposed system is running protein-protein
interaction workflows. However, we plan to use the system in other types of
genomic workflows in the future.
As a conclusion, we believe that most of the attention
in Information Systems has gone to data management and interoperability, and
this attention will increasingly shift to information and knowledge with
support for semantics on the one hand, and processes on the other. The first deals with what is the service and product in
E-commerce applications for example, and the second deals with how to
effectively support or render it. In our vision partly outlined in [Sheth 1999]
and [Sheth, Aalst, and Arpinar 1999], we propose that (a) semantics is the key
to effective organization and utilization of Web-based and Enterprise-based
information regardless of syntax, format/representation, media, and source, (b)
for the processes in future networked economy, in which we will see process as
an organic part of doing a business—that is, while processes will be chief
differentiating and the competitive force in doing business in the networked
economy, they will be deeply integrated with the way of doing business, and
that they will be critical components of almost all types of systems supporting
enterprise-level and business critical activities.
4.
Impact
LSDIS’s research has achieved significant impact
through several avenues, including training of students, commercialization, and
industry collaborations.
Technology Transfer and Commercialization:
§
InfoHarness research lead to its
commercialization as Adapt/X Harness from Bellcore (now Telcordia Technology)
in 1995.
§
METEOR research lead to technology
licensing to Infocosm, Inc., which then released a commercial version called
METEOR EAppS in 1998, and was licensed to enterprises.
§
VideoAnywhere technology resulted
into Venture Capital funded spin-off Taalee, Inc. in 1999 (recently acquired by
Voquette, Inc), which has patented methods related to Semantic Search, Personalization,
Directory and Interactive Marketing, commercialized the Semantic Engine™ technology, and licensed corresponding
commercial products and services [see www.taalee.com and www.voquette.com].
Significant Applications and Trials (a partial list):
§
Medical College of Georgia (MCG) and
LSDIS developed two applications: (a) Pediatric Echocardiograph Consultation
application for use by rural physicians to get consultation at MCG using the
CaTCH system developed at LSDIS (a field trials proved that this Web based
system yielded the same results as those achieved using traditional approach of
viewing tapes on a TV, while replacing overnight tape delivery by instant and
cost-effective Web based delivery of media and patient information), (b)
CareWeb, an advanced Web-based system developed in the Georgia's Family
Connections program to support distance health education and collaboration
among patients and healthcare workers.
§
Infocosm Inc., MCG and Advanced Technology
Institute used LSDIS’s METEOR system to develop workflow applications to
support neonatal clinical pathways.
§
LSDIS developed CAPA (Course Approval
Process Automation) using some of METEOR technology and LSDIS’s expertise in
workflow management. The system has been
operational since 1998. It supports a 30+ step workflow spread over entire UGA
campus with user-base of 130+ departments and 3500+ faculty and staff members,
and has already been used to support approval of over 6000 courses.
§
Connecticut Healthcare Research and
Education Foundation (CHREF) and LSDIS developed State-wide Child Immunization
Tracking application using the METEOR WebWork system
§ METEOR designer has been made available free of charge to instructors and used in conducting advanced graduate courses in this area at several institutions worldwide.
Acknowledgements
Several long term
visitors to the LSDIS lab have participated and enriched our research. These include, Professors Tarcisio Lima (Department of Computer Science, Federal
University of Juiz de Fora, Brazil), Wil van der Aalst (Eindhoven University of Technology), Eduardo
Mena (Universidad de Zaragoza), and Mizuho Iwaihara (
References
D.
Hall, J. Miller, J. Arnold, K. Kochut, A. Sheth and M. Weise, "Using
Workflow to Build an Information Management System for a Geographically
Distributed Genome Sequencing Initiative," Genomics of Plants and Fungi, R.A. Prade and H.J. Bohner, Editors
(2001) Marcel Dekker, Inc.
V. Kashyap and A. Sheth, “Schematic
and Semantic Similarities between Database Objects: A Context-based
Approach,” Very Large Data Bases (VLDB)
Journal, 5(4), October 1996, pp 276-304.
V. Kashyap and A. Sheth, Information
Brokering Across Heterogeneous Digital Data, Kluwer Academic Publishers, August
2000, 248 pages.
K. Kochut, A. Sheth, and J. Miller,
“Optimizing Workflow: Using a CORBA-based, Fully Distributed Process to Create
Scalable, Dynamic Systems,” Component Strategies Journal, March 1999.
the International Journal of Cooperative
Information Systems (IJCIS), 9 (4) (2000), pp 403-425.
J. Miller, D. Palaniswami, A. Sheth,
K. Kochut and H. Singh, "WebWork: METEOR2's Web-Based Workflow
Management System," Journal of Intelligent Information
Systems: Integrating Artificial Intelligence and Database Technologies (JIIS), 10 (2) (March/April 1998) pp.
185-215.
J.
Miller, A. Sheth and K. Kochut, "Perspectives
in Modeling: Simulation, Database and Workflow," Conceptual Modeling: Current Issues and Future
Directions, LNCS-1565, P. Chen, J. Akoka, H. Kangassalo,
B. Thalheim, Eds, (April 1999) pp 154-167. Springer Verlag.
A.
Sheth, K. Shah, K. Parsuraman and S. Mudumbai, “Searching Distributed and
Heterogeneous Digital Media,” 8th IFIP 2.6 Working Conference on
Database Semantics (DS-8), Rotorua, New Zealand, January, 1999.
K. Shah and A. Sheth,
"InfoHarness: An Information Integration Platform for Managing
Distributed, Heterogeneous Information," IEEE Internet Computing,
November-December 1999, pp 18-28.
A. Sheth, Changing Focus in
Interoperability in Information Systems: From System, Syntax, Structure to
Semantics, Interoperating Geographic Information Systems, in Interoperating
Geographic Information System, M. Goodchild, M. Egenhofer, R. Fegeas, and C.
Kottman, Eds, Kluwer Academic Publishers, 1999.
A. Sheth, W. M. P. van der Aalst,
and I. B. Arpinar, Processes Driving the Networked Economy: Process Portals,
Process Vortexes, and Dynamically Trading Processes, IEEE Concurrency Journal,
July-September 1999.
S. Thacker, A. Sheth and S. Patel,
“Complex Relationships for the Semantic Web,” Creating the Semantic Web, D.
Fensel, J. Hendler, H. Liebermann, and W. Wahlster (eds.)
MIT Press, 2001 (in print).
Additional project details and
references can be found at: http://lsdis.cs.uga.edu.
A list of keynotes and invited talks
with associated presentations that further outline our vision is at http://lsids.cs.uga.edu/~amit.