Download RadhaC_paper.doc

.
VIVEKANANDHA COLLEGE OF ENGINEERING
FOR WOMEN
ELAYAMPALAYAM, TIRUCHENGODE, NAMAKKAL DT.
OMNIS/2 - Integrating Digital Libraries and Multimedia
Databases
SUBMITTED BY:
C.RADHA(PRE-FINAL IT )
CONTACT MAIL ID:
[email protected]
1
.
Abstract
Emerging multimedia applications like digital libraries and document
management require new databases. Within the project OMNIS/2 we are integrating
digital library systems and multimedia databases. The resulting system OMNIS/2 will have
a couple of new features compared to today’s digital libraries, like additional storing and
indexing of user-defined multimedia documents, automatic and personal linking concepts,
annotations and personalization. One of our key features will be the linking concept.
Although each full text document in a digital library contains bibliographic references,
which are again stored in the same library system, it is not possible to simply click on them.
Instead a new complex search query needs to be formulated. OMNIS/2 can also be used as
a meta system, enhancing existing catalog or retrieval systems by these features.
Then these legacy applications serve as a document storage layer, while OMNIS/2 forms
the multimedia storage, linking and personalization layer. Thus, even mere retrieval
systems - and today most digital library systems are mere retrieval systems - can be
enriched to interactive multimedia DL-systems by OMNIS/2.
Introduction
A Multimedia Database Management System(MMDBMS) must support
multimedia data types in addition to providing facilities for traditional DBMS functions
like database creation, data modeling, data retrieval, data access and organization, and
dataIndependence. The area and applications have experienced tremendous growth.
Especially with the rapid development of network technology, multimedia database
System gets more tremendous development and multimedia information exchange becomes
very important. This paper reviews the history and the current state-of-the-art in
MMDBMS.
First generation multimedia database systems focused on kernel support
For blobs (binary large objects), to efficiently store the sizeable objects.
The second phase concerned techniques for annotation and linking media
Objects. Most of this activity found itself a breeding ground in user interface research
and multimedia authoring systems. The database merely contains textual
annotations, made accessible efficiently using conventional information retrieval
(IR) techniques. Still, multimedia objects remain non-interpreted with respect to
retrieval.
The third generation of multimedia database retrieval research focuses
On effective techniques for indexing and retrieval by content.
Contribution of this paper
In this paper, we are integrating digital library systems and multimedia databases.
The key scientific questions driving our research are:
2
.
Information access:
How to effectively satisfy a user's information need using the multimedia objects
stored in a database, bridging the gap between the low-level internal representations of
multimedia content and the high-level cognitive processes of the user.
Data management:
How to efficiently derive simple and complex multimedia features for widely
distributed sources of raw material and making this available as an index for query
resolution.
The research contribution of our project lies in the integration of both
aspects in single software architecture, focusing from the start on exibility,
extensibility, as well as efficiency and scalability
Motivation
Existing digital library systems often contain large volumes of data and offer
various retrieval methods ranging from recherché in catalog and meta data (like OPAC) to
complex full text queries (like our previous OMNIS/1 system) or content based search
queries. Quite often it would be a great benefit if the retrieved full text documents would be
again automatically linked for instance by their bibliographic references or meaningful
words. Furthermore, these systems cannot offer the possibility of adding annotations or
personal documents, as they are only retrieval systems, denying any write access of users.
This would be necessary though to move from today’s retrieval systems to interactive
digital library systems, which offer a working environment, i.e. where it is possible to link
digital library documents according to the users’ interests, to annotate the documents, to
create a personalized view on them and to enhance them with your own (multimedia)
documents or already existing ones.
Features of the OMNIS/2 System
Summarizing, OMNIS/2 holds the following new features compared to traditional digital
library systems:
– the integration of full text retrieval library systems and multimedia databases
into a progressive, interactive, multimedia digital library system with the option of
transparently including other systems (with often large collections of data) and even linking
them.
– A generic management of metadata, not only for its own documents, but also for
any document of the additionally connected digital library systems. This is necessary in
order to make use of modern filtering and retrieval techniques and also for the transparent
linking and processing of documents beyond the boundaries of a single digital library
system.
– Automatic linking of documents inside of digital libraries and between documents
of different digital library systems. Every retrieved article from a digital library system
3
.
contains further bibliographic references which are mostly stored in the same digital
library system. To retrieve these references the user usually has to initiate another search
query, possibly with further restrictions, until the correct result is returned.
– User defined, personalized links between documents. While working with
retrieved digital documents, the user wants to follow personal relevant links to other
documents in the same digital library system or to other systems to find relevant content.
Links can be personalized according to single users, user groups, or topics of interest. This
feature is often asked for by research groups who work with special library systems like
VD17.
– Personal annotations. Current library research systems lack the possibility of
adding personal annotations to documents, as this would require write permission for all
users of the digital library system. Annotations are helpful though to explore content for
oneself or for a certain user group or to simply discuss a topic. Discussions require
recursive annotations of variable scope. If various types of media annotations are possible,
a user-friendly authoring tool and the integration of a multimedia database system for
storage purposes are additionally required. Annotations are supported in newly developed
media libraries and personal libraries, but not in catalog and research systems.
– Adaptive personalization. Current digital library systems often miss a
personalization feature which every user can feed with personal interests. This feature can
on one hand be used with an existing pull-technology for an additional semantic filtering of
information (e.g. to put queries into a context in order to eliminate ambiguities of the same
word in different topics) or, on the other hand, as a push-technology to inform a user about
new relevant incoming titles. The personalization feature should be adaptive to recognize
shifts in the users’ interests, but it should also be a corporate tool to emulate
recommendations. We include our GRAS algorithm (Gaussian Rating Adaption Schema)
in OMNIS/2.
The OMNIS/2 System Architecture
OMNIS/2 is an integration of the digital library system OMNIS and the
multimedia database system MultiMAP. The goal is to create a stand-alone, interactive
digital multimedia library system. The full text retrieval capabilities of OMNIS and
the storage capabilities of multimedia documents in MultiMAP including parts of the
database scheme are incorporated into the OMNIS/2 system, which enables the user to
interactively create, store and search for multimedia documents in digital libraries. These
documents can be represented via an interface to the WWW. Documents, consisting of
texts, tables, graphics, audio and video are then not only searchable by queries in
descriptive data (like author, title, keywords, etc.) or by full text queries, but also by a
much more powerful linking concept than the currently existing one in the WWW.
Documents can be annotated with link anchors without changing or rewriting the original
documents. This concept is supported by the Amsterdam Hypermedia Model which is used
in OMNIS/2 as an extension of the Dexter Reference Model. This clearly separates the
document layer (within-component layer) from the linkage
layer (storage layer).
In addition it is desirable to transparently include other existing digital library
systems (catalog systems, full text systems or multimedia systems). The system is modeled
4
.
as a three-tier architecture where the databases are separated from the web server in a
layer of its own. There is no difference in the handling of local documents and the handling
of results from connected external systems. This enables OMNIS/2 to search various other
systems and to automatically link all documents, to annotate them, to extend them with
multimedia components and to personalize them. The original documents themselves
remain in the original database systems and are never modified. They are represented in
the OMNIS/2 database simply by their address and meta data. The linking, including the
anchor positions, is stored in OMNIS/2 exclusively and is included dynamically into the
retrieved documents at run-time. In the same way documents can be annotated with userrelated, group-related or general annotations. These annotations can be multimedia
annotations or even recursive annotations. To create user-defined multimedia documents
or to enhance existing ones, OMNIS/2 is equipped with an easy to use authoring tool. In
addition a concept for a personalization feature (GRAS algorithm) is available, which leads
to a better filtering and a personal ranking of the results of the included digital library
systems. It can also be used as a messaging system about new relevant incoming documents
in the library. Both push and pull strategies are possible. The annotations and the data
necessary for the personalization feature are stored in the OMNIS/2 database. These
features allow two views on OMNIS/2. The ability to integrate various other systems gives
OMNIS/2 the characteristics of a meta system. It is also possible to look at OMNIS/2 as a
stand alone system since it offers features to create, store and search for its own multimedia
documents.
Implementation and Outlook
The system is currently under development. The above is already built as a
demonstrator with an XML coupling to the full text library system .Further improvements
are on their way to turn the demonstrator into a full prototype.
References
Bay95a. R. Bayer. The digital library system omnis/myriad. In Proc. 18th Australasian
Computer Science Conference
(ACSC’95), page 5, Glenelg, Adelaide, South Australia, 1995.
Bay95b. R. Bayer. Document management as a database problem. In Proc. of the 21st Int.
Conference on Very Large Data
Bases (VLDB’95), pages 7–10, Zürich, 1995.
BKH+99. R. Bayer, R. Kallenborn, H. Haddouti, W. Wohner, A. Mödl, R. Heinrich, and D.
Nitsche. Integration of digital
services for libraries. In First Russian National Conference on Digital Libraries: Advanced
Methods And Technologies,
Digital Collections., pages 46–51, Saint Petersburg, Russia, October 1999.
BOWV95. C. Böhm, A. Oppitz, S. Wiesener, and P. Vogel. Prints of the 17th century in a
distributed digital library system.
In Norman Revell and A. Min Toja, editors, Proc. of the 6th Int. Conference on Database
and Expert Systems Applicationss (DEXA’95), London, 1995.
5
.
6